Electrochemical, Spectroscopic, and DFT Study of C60(CF3)n Frontier

Aug 24, 2007 - The 18 experimental 0/− E1/2 values (vs C600/-) are a linear ... Citation data is made available by participants in Crossref's Cited-...
0 downloads 0 Views 745KB Size
Published on Web 08/24/2007

Electrochemical, Spectroscopic, and DFT Study of C60(CF3)n Frontier Orbitals (n ) 2-18): The Link between Double Bonds in Pentagons and Reduction Potentials Alexey A. Popov,*,†,‡ Ivan E. Kareev,§,| Natalia B. Shustova,⊥ Evgeny B. Stukalin,O Sergey F. Lebedkin,| Konrad Seppelt,# Steven H. Strauss,*,⊥ Olga V. Boltalina,*,⊥ and Lothar Dunsch*,‡ Contribution from the Chemistry Department, Moscow State UniVersity, Moscow, 119992, Russia, Department of Electrochemistry and Conducting Polymers, Leibniz Institute for Solid State and Materials Research Dresden, Helmholtzstrasse 20, Dresden D01069, Germany, Institute of Problems of Chemical Physics, Russian Academy of Sciences, ChernogoloVka 142432, Russia, Forschungszentrum Karlsruhe, Institute for Nanotechnology, Karlsruhe 76021, Germany, Department of Chemistry, Colorado State UniVersity, Fort Collins, Colorado 80523, The James Franck Institute, UniVersity of Chicago, Chicago, Illinois 60637, and Institute for Inorganic and Analytical Chemistry, Free UniVersity Berlin, D14195 Berlin, Germany Received May 17, 2007; E-mail: [email protected]; [email protected]; [email protected]; [email protected]

Abstract: The frontier orbitals of 22 isolated and characterized C60(CF3)n derivatives, including seven reported here for the first time, have been investigated by electronic spectroscopy (n ) 2 [1], 4 [1], 6 [2], 8 [5], 10 [6], 12 [3]; the number of isomers for each composition is shown in square brackets) fluorescence spectroscopy (n ) 10 [4]), cyclic voltammetry under air-free conditions (all compounds with n e 12), ESR spectroscopy of C60(CF3)n- radical anions at 25 °C (n ) 4 [1] and 10 [1]), and quantum chemical calculations at the DFT level of theory (all compounds including n ) 16 [3] and 18 [2]). DFT calculations are also reported for several hypothetical C60(CF3)n derivatives. The X-ray structure of one of the compounds, 1,6,11,16,18,26,36,41,44,57-C60(CF3)10, is reported here for the first time. Most of the compounds with n e 12 exhibit two or three quasi-reversible reductions at scan rates from 20 mV s-1 up to 5.0 V s-1, respectively. The 18 experimental 0/- E1/2 values (vs C600/-) are a linear function of the DFT-predicted LUMO energies (average E1/2 deviation from the least-squares line is 0.02 V). This linear relationship was used to predict the 0/- E1/2 values for the n ) 16 and 18 derivatives, and none of the predicted values is more positive than the 0/- E1/2 value for one of the isomers of C60(CF3)10. In general, reduction potentials for the 0/couple are shifted anodically relative to the C600/- couple. However, the 0/- E1/2 values for a given composition are strongly dependent on the addition pattern of the CF3 groups. In addition, LUMO energies for isomers of C60(X)n (n ) 2, 4, 6, 8, 10, and 12) that are structurally related to many of the CF3 derivatives were calculated and compared for X ) CH3, H, Ph, NH2, CH2F, CHF2, F, NO2, and CN. The experimental and computational results for the C60(CF3)n compounds and the computational results for more than 50 additional C60(X)n compounds provide new insights about the frontier orbitals of C60(X)n derivatives. For a given substituent, X, the addition pattern is as important, if not more important in many cases, than the number of substituents, n, in determining E1/2 values. Those addition patterns with double bonds in pentagons having two C(sp2) nearest neighbors result in the strongest electron acceptors.

Introduction

The observation that fullerenes and their derivatives are powerful electron acceptors has attracted considerable attention because of the potential applications of these compounds to problems in energy storage and photovoltaic energy conversion.1-3

The gas-phase electron affinities of C60 (g2.666(1) eV),4,5 C70 (2.676(1) eV),4 and C76-D2(1) (2.88(5) eV)6 rival those of atomic sulfur (2.07 eV) and chlorine (3.61 eV). Multiply charged ions such as C604- are kinetically stable in the gas phase,7 and up to six reversible one-electron reductions have been observed in some solvents.8,9



Moscow State University. Leibniz Institute for Solid State and Materials Research Dresden. Russian Academy of Sciences. | Institute for Nanotechnology. ⊥ Colorado State University. O University of Chicago. # Free University Berlin. ‡ §

(1) Nierengarten, J.-F. New J. Chem. 2004, 28, 1177-1191. (2) Hoppe, H.; Sariciftci, N. S. J. Mater. Res. 2004, 19, 1924-1945. 10.1021/ja073181e CCC: $37.00 © 2007 American Chemical Society

(3) Segura, J. L.; Martin, N.; Guldi, D. M. Chem. Soc. ReV. 2005, 34, 31-47. (4) Brink, C.; Andersen, L. H.; Hvelplund, P.; Mathur, D.; Voldstad, J. D. Chem. Phys. Lett. 1995, 233, 52-54. (5) Wang, X.-B.; Ding, C.-F.; Wang, L.-S. J. Chem. Phys. 1999, 110, 82178220. (6) Boltalina, O. V.; Sidorov, L. N.; Borchshevskii, A. Y.; Sukhanova, E. V.; Skokan, E. V. Rapid Commun. Mass. Spectrom. 1993, 7, 1009-1011. (7) Cammarata, V.; Guo, T.; Illies, A.; Li, L.; Shevlin, P. J. Phys. Chem. A 2005, 109, 2765-2767. J. AM. CHEM. SOC. 2007, 129, 11551-11568

9

11551

Popov et al.

ARTICLES

The electron-acceptor properties of fullerene(X)n derivatives, as measured by electrochemical E1/2 values, are believed to be modulated primarily by two well-known factors:10-14 (i) saturation of the fullerene π system (i.e., the value of n), which raises the LUMO energy and makes the first reduction less favorable relative to the parent fullerene; and (ii) the electron-releasing or electron-withdrawing nature of the n substituents, which can augment or diminish the effect of π-system saturation on E1/2. One of our goals in fullerene science and technology is to prepare and study a series of fullerene(X)n derivatives with as wide a range of electron-accepting properties as possible.15 This will lead to a better understanding of how different X groups, different values of n (for a given X), and different addition patterns (for a given X and a given value of n) affect electrochemical potentials and anion-radical lifetimes (the latter affects the reversibility of the electron-transfer process). Furthermore, a systematic investigation of this type provides detailed information about the nature of the frontier orbitals of fullerene derivatives, and therefore about the reactivity of the derivatives, information that synthetic chemists will be able to use in their pursuit of regioselective addition methodologies. It is now the case that there are a greater number of isolated and well-characterized homo-addended fullerene(X)n compounds for X ) CF3 than for any other substituent X that forms a 2c2e- bond with one cage C atom (for example, there are now 28 fullerene(CF3)n X-ray structures and more than 20 additional compounds that have been characterized by other physicochemical methods; see Table S-1 in Supporting Information (SI) for a complete list of references). Among these are 24 C60(CF3)n derivatives, which are listed in Table 1.16-37 (Here and (8) Echegoyen, L.; Echegoyen, L. E. Acc. Chem. Res. 1998, 31, 593-601. (9) Herranz, M. A.; Echegoyen, L. In Fullerenes: From Synthesis to Optoelectronic Properties; Guldi, D. M., Martin, N., Eds.; Kluwer Academic: Dordrecht, The Netherlands, 2002; pp 267-293. (10) Guldi, D.; Hungerbuhler, H.; Asmus, K. D. J. Phys. Chem. 1995, 99, 93809385. (11) Boudon, C.; Gisselbrecht, J. P.; Gross, M.; Isaacs, L.; Anderson, H. L.; Faust, R.; Diederich, F. HelV. Chim. Acta 1995, 78, 1334-1344. (12) Echegoyen, L.; Diederich, F.; Echegoyen, L. E. In Fullerenes: Chemistry, Physics, and Technology; Kadish, K., Ruoff, R., Eds.; Wiley-Interscience: New York, 2000; pp 1-52. (13) Martin, N.; Sanchez, L.; Illescas, B.; Perez, I. Chem. ReV. 1998, 98, 25272547. (14) Suzuki, T.; Maruyama, Y.; Akasaka, T.; Ando, W.; Kobayashi, K.; Nagase, S. J. Am. Chem. Soc. 1994, 116, 1359-1363. (15) Popov, A. A.; Tarabek, J.; Kareev, I. E.; Lebedkin, S. F.; Strauss, S. H.; Boltalina, O. V.; Dunsch, L. J. Phys. Chem. A 2005, 109, 9709-9711. (16) Avent, A. G.; Boltalina, O. V.; Goryunkov, A. V.; Darwish, A. D.; Markov, V. Y.; Taylor, R. Fullerenes Nanotubes Carbon Nanostruct. 2002, 10, 235241. (17) Darwish, A. D.; Abdul-Sada, A. K.; Avent, A. G.; Lyakhovetsky, V. I.; Shilova, E. A.; Taylor, R. Org. Biomol. Chem. 2003, 1, 3102-3110. (18) Darwish, A. D.; Avent, A. G.; Abdul-Sada, A. K.; Taylor, R. Chem. Commun. 2003, 1374-1375. (19) Goryunkov, A. A.; Kuvychko, I. V.; Ioffe, I. N.; Dick, D. L.; Sidorov, L. N.; Strauss, S. H.; Boltalina, O. V. J. Fluorine Chem. 2003, 124, 61-64. (20) Goryunkov, A. A.; Ioffe, I. N.; Kuvychko, I. V.; Yankova, T. S.; Markov, V. Y.; Streletskii, A. V.; Dick, D. L.; Sidorov, L. N.; Boltalina, O. V.; Strauss, S. H. Fullerenes Nanotubes Carbon Nanostruct. 2004, 12, 181185. (21) Kareev, I. E.; Shustova, N. B.; Kuvychko, I. V.; Lebedkin, S. F.; Miller, S. M.; Anderson, O. P.; Popov, A. A.; Strauss, S. H.; Boltalina, O. V. J. Am. Chem. Soc. 2006, 128, 12268-12280. (22) Kareev, I. E.; Kuvychko, I. V.; Lebedkin, S. F.; Miller, S. M.; Anderson, O. P.; Strauss, S. H.; Boltalina, O. V. Chem. Commun. 2006, 308-310. (23) Popov, A. A.; Kareev, I. E.; Shustova, N. B.; Peryshkov, D. V.; Miller, S. M.; Anderson, O. P.; Boltalina, O. V.; Strauss, S. H. 2007. Unpublished X-ray structure of 60-12-3 (some disorder; possibly a mixture of two C2 and one C2h isomers, which have nearly identical DFT-predicted relative energies and DFT-predicted LUMO energies). (24) Kareev, I. E.; Shustova, N. B.; Newell, B. S.; Miller, S. M.; Anderson, O. P.; Strauss, S. H.; Boltalina, O. V. Acta Crystallogr. 2006, E62, o3154o3156. 11552 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

elsewhere, C60(CF3)n compounds that have been isolated and characterized will be denoted as 2-1, 4-1, etc. The abbreviation for hypothetical C60(CF3)n compounds will always carry the suffix -CF3 for clarity. In addition, abbreviations for C60(X)n compounds with X other than CF3, real or hypothetical, will always carry the suffix -X.) The addition patterns for 18 of them are shown as Schlegel diagrams in Figure 1. The IUPAC numbering for C60 and for 16 of the 18 compounds studied in this work are shown in Figures S-1 and S-2, respectively. Five of the other addition patterns are shown in Figure S-3 (see SI). The compound 18-2-CF3 has not yet been isolated but is also included in Table 1 because it is believed to be the most stable isomer of this composition.32 The structures of nine of the 18 compounds with n e 12 are known from X-ray crystallography (8-1, 8-2, 10-2, 10-3, 10-4, 10-5, 10-6, 12-1, 12-2, and 12-3), and the structures of four others were elucidated previously by a combination of 1D and 2D 19F NMR spectroscopy and DFT calculations (2-1, 4-1, 6-1, and 6-2). Five of the 18 isolated compounds with n e 12 are reported here for the first time (8-3, 8-4, 8-5, 10-5, and 12-3). Also listed in Table 1 are two addition patterns known from the X-ray structures of C60(C2F5)6 and C60(C2F5)8 and one addition pattern known from the X-ray structure of C60(CMe(COOCMe3)2)6. Schlegel diagrams for these three derivatives are shown in Figure S-4 (see SI). In this paper we report a combination of electrochemical data, electronic, fluorescence, and ESR spectra, and DFT calculations for the 22 known C60(CF3)n derivatives and several hypothetical C60(CF3)n derivatives (four of the 0/- E1/2 values were briefly reported in 200515). We also report DFT calculations for known or hypothetical C60(X)n compounds for many of the addition patterns shown in Figure 1 for X ) H, Ph, CH3, CH2F, CHF2, NH2, NO2, CN. In addition to their fundamental importance, these results suggest strategies for the design of new exohedral fullerene derivatives with an even wider range of electronic properties than is currently available. (25) Kareev, I. E.; Kuvychko, I. V.; Lebedkin, S. F.; Miller, S. M.; Anderson, O. P.; Seppelt, K.; Strauss, S. H.; Boltalina, O. V. J. Am. Chem. Soc. 2005, 127, 8362-8375. (26) Kareev, I. E.; Lebedkin, S. F.; Miller, S. M.; Anderson, O. P.; Strauss, S. H.; Boltalina, O. V. Acta Crystallogr. 2006, E62, o1498-o1500. The lowest locants were incorrectly given in this paper as 1,6,11,16,18,26,36,44,48,58. They should be 1,6,11,16,18,24,27,36,54,60. (27) Kareev, I. E.; Lebedkin, S. F.; Popov, A. A.; Miller, S. M.; Anderson, O. P.; Strauss, S. H.; Boltalina, O. V. Acta Crystallogr. 2006, E62, o1501o1503. (28) Kareev, I. E.; Shustova, N. B.; Peryshkov, D. V.; Lebedkin, S. F.; Miller, S. M.; Anderson, O. P.; Popov, A. A.; Boltalina, O. V.; Strauss, S. H. Chem. Commun. 2007, 1650-1652. (29) Troyanov, S. I.; Dimitrov, A.; Kemnitz, E. Angew. Chem., Int. Ed. 2006, 45, 1971-1974. (30) Shustova, N. B.; Kuvychko, I. V.; Bolskar, R. D.; Seppelt, K.; Strauss, S. H.; Popov, A. A.; Boltalina, O. V. J. Am. Chem. Soc. 2006, 128, 1579315798. (31) Dorozhkin, E. I.; Ignat’eva, D. V.; Tamm, N. B.; Vasilyuk, N. V.; Goryunkov, A. A.; Avdoshenko, S. M.; Ioffe, I. N.; Sidorov, L. N.; Pattison, P.; Kemnitz, E.; Troyanov, S. I. J. Fluorine Chem. 2006, 127, 13441348. (32) Troyanov, S. I.; Goryunkov, A. A.; Dorozhkin, E. I.; Ignat’eva, D. V.; Tamm, N. B.; Avdoshenko, S. M.; Ioffe, I. N.; Markov, V. Y.; Sidorov, L. N.; Scheural, K.; Kemnitz, E. J. Fluorine Chem. 2007, 128, 545551. (33) Canteenwala, T.; Padmawar, P. A.; Chiang, L. Y. J. Am. Chem. Soc. 2005, 127, 26-27. (34) Popov, A. A.; Senyavin, V. M.; Troyanov, S. I. J. Phys. Chem. A 2006, 110, 7414-7421. (35) Shustova, N. B.; Peryshkov, D. V.; Popov, A. A.; Boltalina, O. V.; Strauss, S. H. Acta Crystallogr. 2007, E62, o3129. (36) Goryunkov, A. A.; Dorozhkin, E. I.; Tamm, N. B.; Ignat’eva, D. V.; Avdoshenko, S. M.; Sidorov, L. N.; Troyanov, S. I. MendeleeV Commun. 2007, 17, 110-112. (37) Shustova, N. B.; Peryshkov, D. V.; Kareev, I. E.; Boltalina, O. V.; Strauss, S. H. Acta Crystallogr. 2007, E63, o3398.

Investigation of Frontier Orbitals of Fullerene Derivatives

ARTICLES

Table 1. C60(Rf)n Derivatives (n ) 2, 4, 6, 8, 10, 12, 16, 18)a synthesis and characterization references addition pattern

cmpd

1,7-C60(CF3)2 1,6,11,18-C60(CF3)4 1,6,11,18,24,27-C60(CF3)6 1,6,9,12,15,18-C60(CF3)6 1,7,16,36,46,49-C60(C2F5)6 1,23,28,33,38,60-C60(CF3)6d 1,6,11,16,18,24,27,36-C60(CF3)8 1,6,11,18,24,27,52,55-C60(CF3)8 1,6,11,18,24,27,53,56-C60(CF3)8e 1,6,11,16,18,28,31,36-C60(CF3)8e 1,6,11,18,24,27,33,51-C60(CF3)8e,f 1,6,11,18,24,27,32,35-C60(C2F5)8 1,6,11,16,18,24,27,36,41,57-C60(CF3)10e 1,6,11,16,18,26,36,44,48,58-C60(CF3)10 1,3,7,10,14,17,23,28,31,40-C60(CF3)10 1,6,12,15,18,23,25,41,45,57-C60(CF3)10 1,6,11,16,18,26,36,41,44,57-C60(CF3)10 1,6,11,18,24,27,33,51,54,60-C60(CF3)12 1,6,11,16,18,26,36,44,46,49,54,60-C60(CF3)12 1,3,6,11,13,18,26,32,35,41,44,57-C60(CF3)12 1,6,9,12,15,18,43,46,49,52,55,60-C60(CF3)12 1,3,6,11,13,18,22,28,31,34,37,41,43,46,51,59-C60(CF3)16h 1,3,6,8,11,13,18,23,28,31,34,35,37,50,54,60-C60(CF3)16h 1,3,6,11,13,18,22,24,27,33,41,43,46,49,51,59-C60(CF3)16h 1,3,6,8,11,13,18,23,28,31,34,37,43,46,51,53,56,59-C60(CF3)18h 1,3,6,11,13,18,22,24,27,32,35,37,41,43,46,49,52,54-C60(CF3)18h

Cs-para (p) C1-pmp C1-p3mp Cs-SPPc C1-p,p,p D3d-(2,6-C10(CF3)2)6-loop C1-p3mpmp C1-p3mp,p C1-p3mpmp C1-pmpmpmp C1-p3mp,p C1-p3mp,p C1-p3mpmp,p C1-p3mpmpmp C1-pmp3mpmp C2-(p3m2-loop)2 C1-pmpmpmpmp C1-p3mpmp,pmp S6-(pm)6-loop C1-p3mpmpmpmp C2(h)-(2 × SPP)c

abbreviation

2-1 4-1 6-1 6-2 6-3-C2F5 6-4-CF3 8-1 8-2 8-3 8-4 8-5 8-6-C2F5 10-1 10-2 10-3 10-4 10-5 10-6 12-1 12-2 12-3g 16-1 16-2 16-3 18-1 18-2-CF3

X-ray

b 22 33d 36,37 24

22 26 25 27 tw 35 29 28 23 32 32 32 32

19

F NMR

DFT

16-19 19 19 21 22

20, tw 20, tw 20, tw 21, tw 22, tw 34, tw tw tw tw tw tw 22, tw tw 26, tw tw 27, tw tw tw 29,30, tw 28, tw tw 32, tw 32, tw 32, tw 32, tw 32, tw

tw tw tw tw tw 22 25 25 25 tw tw tw 30 28 tw

a tw ) this work. b This addition pattern has been observed on the C -like pole of C in C (CF ) . c SPP ) skew pentagonal-pyramid addition pattern. 60 70 70 3 10 This is a hypothetical compound based on the X-ray structure of the known compound C60(CMe(CO2CMe3)6. e This is the most probable addition pattern, but is considered tentative until proven by X-ray crystallography. f An equally probable addition pattern differs only in the placement of the isolated hexagon: 1,6,11,18,24,27,51,59-C60(CF3)8. g This product may be a mixture of up to three isomers (two C2 isomers and one C2h isomer, all with two skew-pentagonalpyramidal arrays of six CF3 groups on opposite C60 poles. The locants shown here are for the C2h isomer. h See Supporting Information for the complex addition patterns of these five derivatives.

d

Experimental Section Reagents, Solvents, and Previously Reported C60(CF3)n Derivatives. The reagents and solvents CF3I (Apollo Scientific), C60 (Term USA), hexafluorobenzene (Sigma-Aldrich), 1,2-dichlorobenzene (SigmaAldrich, 99%, anhydrous), ferrocene (Fe(Cp)2, Fluka), decamethylferrocene (Fe(Cp*)2, Fluka), cobaltocene (Co(Cp)2, Fluka), chloroform-d and benzene-d6 (Cambridge Isotopes), and heptane and toluene for HPLC purification (Sigma-Aldrich) were used as received. The following solvents and supporting electrolyte for cyclic voltammetry were purified as indicated: dichloromethane (Fluka, puriss. grade; stored in a nitrogen-filled glovebox over 4 Å molecular sieves (Merck)); 1,2dichlorobenzene (Sigma-Aldrich, ACS grade; distilled from CaH2 and stored in the glovebox); tetrabutylammonium tetrafluoroborate (N(nBu)4BF4, Fluka puriss. grade, stored in the glovebox after drying under vacuum at 70 °C for 24 h). The following C60(CF3)n derivatives were prepared as previously described: 2-1, 4-1, and 6-1;19 6-2;21 8-2;24 101, 10-2, and 10-3;25 10-4;27 12-1;28,29 and 12-2.28 Synthesis and Isolation of 8-1, 8-3, 8-4, 8-5, 10-5, 10-6, and 12-3. In a typical experiment, finely ground C60 (150 mg, 0.208 mmol) was placed in a 0.8-cm I.D. glass tube connected to a gas handling system at one end and a mineral-oil bubbler at the other. The portion of the tube containing C60 was placed in a 5-cm long tube furnace. After purging the sample with argon, it was heated to 460 °C and treated with 12 sccm of gaseous CF3I (ca. 0.05 mmol min-1) for 2 h [CAUTION: CF3I decomposes in air above 300 °C and produces toxic HF, COF2, and I2; handle only in a well-ventilated fume hood]. Orangebrown C60(CF3)n derivatives and purple I2 condensed inside the tube approximately 1 cm outside of both ends of the furnace (i.e., in the cold zones). Iodine was removed under a flow of argon with mild heating (e100 °C).

The orange-brown condensate (a total of 600 mg from several preps) was purified by HPLC as follows. In the first stage of purification (20mm I.D. × 250 mm Cosmosil Buckyprep column, Nacalai Tesque, Inc.; 300 nm UV detector; 18 mL min-1 eluent flow rate), 1.8-mL injections were eluted with toluene to give eight fractions. In the second stage, 1.8 mL-injections of each fraction were eluted with 20/80 (v/v) toluene/hexane. One first-stage fraction contained several isomers of C60(CF3)12, one of which (60-12-3) was isolated to 98+% purity using 100% heptane as eluent (its retention time on a 10-mm I.D. × 250 mm Buckyprep column was 5.9 min at 3 mL/min flow rate). One second-stage fraction contained primarily two isomers, 10-1 and 10-2, which were separated as narrow cuts and isolated by solvent evaporation (80+ mol % purity). They were further purified to 90% and 98%, respectively, by a third HPLC stage using 100% heptane (10-mm I.D. × 250 mm Cosmosil Buckyprep column, Nacalai Tesque, Inc.; 300nm UV detector; 5 mL min-1 flow rate, retention times 10.4 min for 10-1 and 14.7 min for 10-2). Another second-stage fraction contained primarily 10-3 and smaller amounts of 10-4 and 10-5. A third stage of HPLC purification of this fraction (same conditions as second stage) using 20/80 toluene/hexane consisted of collecting a cut from 15.2 to 16.8 min. This contained 10-3 with 95+% purity. The compounds 10-4 and 10-5 had 20/80 toluene/hexane retention times of 11.2 and 7.3 min, respectively, and were also isolated with 95+% purity. Compound 6010-6 (95+% purity) was isolated from a 500 °C hot-tube reaction product using a 10-mm I.D. × 250 mm Buckyprep column. Its retention time in 100% heptane at 3 mL/min flow rate was 21 min. The five isomers of C60(CF3)8 were isolated from the 4.5-6.5 min first-stage fraction. Second-stage purification using 20/80 toluene/ heptane resulted in the collection of 8-1, 8-2, 8-3, 8-4, and 8-5 at 13.5, 14.5, 16.0, 18.8, and 19 min, respectively. Two more stages of HPLC J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11553

Popov et al.

ARTICLES

Figure 1. Schlegel diagrams for the 18 C60(CF3)n compounds investigated with n e 12. The black circles indicate the cage carbon atoms to which the CF3 groups are attached. The isolated p-C6(CF3)2 hexagons and the ribbons and loops of edge-sharing m- and p-C6(CF3)2 hexagons are highlighted, and the m-C6(CF3)2 hexagons are indicated with the letter m. The diagrams have been drawn to show similarities in parts of their addition patterns, not to depict the lowest locants relative to a fixed numbering scheme. IUPAC lowest-locant Schlegel diagrams for all but two of the compounds investigated are shown in the Supporting Information. The bonds highlighted in red are those double bonds in pentagons that have two C(sp2) nearest neighbors.

purification under the same conditions resulted in 90+% pure samples of these compounds except for 8-2, which was contaminated with ca. 40% of 8-3 (confirmed by 19F NMR spectroscopy; see Figure S-5 in SI). Physicochemical Measurements. Cyclic voltammetry experiments were carried out in the glovebox (water and oxygen content below 1 ppm) in a one-compartment electrochemical cell. The electrolyte solution was 0.1 M N(n-Bu)4BF4 in dichloromethane except for 12-1, which for solubility reasons was 0.1 M N(n-Bu)4BF4 in 1,2-dichlorobenzene. Control experiments with several n < 12 derivatives demonstrated that the potentials in both solvents relative to C60 potentials were within ( 0.01 V, which is the same as the experimental error in measured potentials. The working electrode was a platinum wire terminated with a platinum plate (0.04 cm2) for n < 12 except for 10-6 and was a glassy-carbon electrode for 10-6, 12-1, 12-2, and 12-3. For all compounds, a platinum wire loop and a silver wire served as the counter electrode and quasi-reference electrode, respectively. The potentials were measured relative to the Fe(Cp*)2+/0 or Fe(Cp)2+/0 potentials (i.e., either Fe(Cp*)2 or Fe(Cp)2 was added as an internal 11554 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

standard). The experiments were controlled by a PAR 263 or 273A potentiostat/galvanostat. The ESR spectrum of the radical anion 4-1was recorded using a spectroelectrochemical cell with a laminated platinum-mesh working electrode (1024 meshes/cm2; 0.08 cm2 active surface area; silver wire quasi-reference electrode (see above)). The airtight cell was filled with a dichloromethane solution of 4-1 and 0.1 M N(n-Bu)4BF4 in the glove box and transferred to the Bruker ER4104OR optical cavity of a Bruker X-band EPR spectrometer. The potential was held at -0.1 V vs C600/- and the ESR spectrum was recorded. The ESR spectrum of the 10-2- radical anion was recorded with the same instrumentation. The sample was a 1,2-dichloromethane solution of 10-2 to which 1 equiv of Co(Cp)2 had been added. Electronic spectra of dichloromethane and/or toluene solutions of the C60(CF3)n compounds were recorded using a Shimadzu 3100 or a Varian Cary 500 spectrophotometer. Emission spectra of toluene solutions were recorded using a Spex Fluorolog-3 spectrometer. The emission was detected in a 90° geometry using a Hamamatsu R928 photomultiplier sensitive in the visible region to ca. 850 nm. Emission spectra were corrected for the wavelength-dependent instrument response. Fluorine19 NMR spectra of chloroform-d or benzene-d6 solutions containing a small amount of hexafluorobenzene as an internal standard (δ -164.9) were recorded at 25 °C using a Varian INOVA-unity 400 spectrometer operating at 367.45 MHz. DFT Calculations. Geometry optimization of all structures in both neutral and anionic forms was done in vacuo with the use of the PBE functional38 and the TZ2P-quality basis set implemented in PRIRODA package.39 Evaluation of Coulombic and exchange-correlation terms was accelerated by expansion of the electron density in an auxiliary basis set.39 The molecular geometry optimization using this methodology typically took two or 3 days with one Opteron CPU (in contrast, molecular geometry optimization with hybrid functionals such as B3LYP typically takes weeks). Single-point energy calculations at the B3LYP/6-311G* level in vacuo and in dichloromethane were carried out using the Gaussian 03 package (these calculations typically took several days for each C60(CF3)n molecule or C60(CF3)nm- ion with one Opteron CPU).40 Electrostatic contributions to the solvation energy were evaluated using the conductor-like polarizable continuum model (C-PCM)41,42 as implemented in Gaussian 03. To construct the cavity encapsulating the solute, we used PBE0/6-31G*-optimized atomic radii (UAKS method in Gaussian 03) because they provide the most balanced description for both neutral and charged solutes when coupled to DFT computations.43 To avoid artificial solvation inside the fullerene cage, an additional sphere with radius 3.5 Å placed at the center of C60 was added to the cavity. X-ray Diffraction. Crystals of 10-5 were grown by slow evaporation from a saturated p-xylene solution at 23 °C. X-ray diffraction data were recorded on a Bruker Smart CCD 1000 TU diffractometer employing Mo KR radiation (graphite monochromator), a scan width of 0.3° in ω, and a measuring time of 40 s/frame, obtaining a full shell of 1800 frames up to 2θ ) 54.0°. Selected details related to this crystallographic experiment are listed in Table 2. The structure was solved by using direct methods and refined (on F2, using all data) by a full-matrix, weighted least-squares process. Absorption and other corrections were applied by using SADABS.44 All non-hydrogen atoms were refined by using anisotropic displacement parameters. Hydrogen atoms were placed in idealized positions and (38) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 38653868. (39) Laikov, D. N. Chem. Phys. Lett. 1997, 281, 151-156. (40) Frisch, M. J.; et al. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (41) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995-2001. (42) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003, 24, 669-681. (43) Takano, Y.; Houk, K. N. J. Chem. Theory Comput. 2005, 1, 70-77. (44) Sheldrick, G. M. SADABS: A program for area detector absorption corrections. Bruker AXS Inc.: Madison, 2003.

Investigation of Frontier Orbitals of Fullerene Derivatives

ARTICLES

Table 2. Crystal Data and Structure Refinements for C1-pmpmpmpmp-C60(CF3)10 (10-5)

cmpd molecular formula formula weight crystal system, space group, Z color of crystal unit cell dimensions a (Å) b (Å) c (Å) R (deg) β (deg) γ (deg) temperature (K) final R indicesa [I > 2σ(I)] goodness-of-fit on F2 a

1,6,11,16,18,26,36,41,44,57-C60(CF3)10‚p-C6H4(CH3)2 C78H10F30 1516.86 g mol-1 triclinic, P1h, 2 orange 12.310(3) 14.197(3) 16.002(3) 90.100(5) 107.630(5) 98.136(5) 173(2) R1 ) 0.0507, wR2 ) 0.1129 1.010

R1 ) (∑||Fo| - |Fc||)/∑ |Fo|; wR2 ) (∑[w(Fo2 - Fc2)2]/∑[w(Fo2)2])1/2.

refined by using a riding model. Bruker APEX2 software was employed for data collection and reduction, and Bruker SHELXTL45 software was used for structure solution, refinement, and graphics.

Results

NMR Spectroscopy. Fluorine-19 NMR spectra of 8-1, 8-2, 8-3, 8-4, 8-5, 10-4, 10-5, 10-6, and 12-3 are reported here for the first time and are shown in Figure S-5 (see SI). The purity of these nine derivatives is greater than 90% except for the sample of 8-2, which also contained ca. 40% of 8-3, and the sample of 8-5, which contained ca. 20% of an unidentified isomer of C60(CF3)8. All nine spectra consist of the appropriate number of multiplets that are either quartets, quartets of quartets, or (in the sole case of 12-3) quartets of quartets of quartets. Some of the latter two are apparent septets or apparent dectets, respectively, when the two JFF values are similar. The timeaveraged 7JFF values for the quartets range from 11.3 to 14.7 Hz (the slow-exchange values, which were not observed in this study, would range from 102(2) to 132(2) Hz; the precision of the time-averaged JFF values in this study is (0.2 Hz). The JFF coupling in fullerene(CF3)n compounds is primarily due to direct lone pair-lone pair interactions between proximate F atoms on CF3 groups that share the same hexagon (i.e., “through-space” Fermi-contact coupling).21,25,30,46-50 The 19F δ and JFF values reported here are similar to those reported for the other nine characterized C60(CF3)n compounds with n e 12, indicating that the CF3 addition patterns of 8-1, 8-2, 8-3, 8-4, 8-5, 10-4, 10-5, 10-6, and 12-3 are almost certainly combinations of isolated p-C6(CF3)2 hexagons and/or ribbons or loops of edge-sharing m- and p-C6(CF3)2 hexagons on the surface of C60 (each shared edge has one sp3 carbon atom and therefore one CF3 group).19,21,25,28,30 This was proven recently by X-ray crystallography for 8-1, 8-2, 10-2, 10-3, 10-4, 10-6, 12-1, 12-2, and 12-3 (see Table 1) and was similarly proven in the current study for 10-5 (see below). Alternative addition patterns for 8-3, 8-4, and 8-5 with CF3 groups on adjacent cage C atoms, as in 6-2,21 (45) Sheldrick, G. M. SHELXTL, v. 6.14; Bruker AXS: Madison, WI 2004. (46) Kareev, I. E.; Santiso-Quinones, G.; Kuvychko, I. V.; Ioffe, I. N.; Goldt, I. V.; Lebedkin, S. F.; Seppelt, K.; Strauss, S. H.; Boltalina, O. V. J. Am. Chem. Soc. 2005, 127, 11497-11504. (47) Alkorta, I.; Elguero, J. E. Struct. Chem. 2004, 15, 117-120. (48) Ernst, L.; Ibrom, K. Angew. Chem., Int. Ed. 1995, 34, 1881-1882. (49) Arnold, W. D.; Mao, J.; Sun, H.; Oldfield, E. J. Am. Chem. Soc. 2000, 122, 12164-12168. (50) Dorozhkin, E. I.; Ignat’eva, D. V.; Tamm, N. B.; Goryunkov, A. A.; Khavrel, P. A.; Ioffe, I. N.; Popov, A. A.; Kuvychko, I. V.; Streletskiy, A. V.; Markov, V. Y.; Spandl, J.; Strauss, S. H.; Boltalina, O. V. Chem. Eur. J. 2006, 12, 3876-3889.

or with one or more 1,3-C5(CF3)2 pentagons, as in 10-325and 12-2,28 can be ruled out because these structural units have unique NMR signatures that are absent from the spectra of 8-3, 8-4, and 8-5. The 19F NMR spectra of 8-2, 8-3, and 8-5 each include four quartets, two of which have the same JFF value (12.4 Hz for 8-2, 12.0 Hz for 8-3 and 11.5 Hz for 8-5), indicative of a ribbon of six CF3 groups plus an isolated p-C6(CF3)2 hexagon and overall C1 symmetry. The 7JFF coupling constants for the terminal CF3 quartets on the ribbon in each compound, which are 12.0 and 13.2 Hz for 8-2 and 11.3 and 14.7 Hz for both 8-3 and 8-5, are similar to the 7JFF values for the CF3 groups at the termini of the p3mp ribbon in 6-1, 11.4 and 14.0 Hz.19 This strongly suggests that the ribbons of six CF3 groups in these two isomers of C60(CF3)8 are p3mp ribbons, not pmpmp ribbons. In the case of 8-2, this was proven by X-ray crystallography.24 The particular p3mp,p isomers that we have assigned to 8-3 and 8-5, which are shown as Schlegel diagrams in Figure 1, are based on an analysis of their 19F NMR spectra (see Figure S-5) and on their DFT-predicted relative energies (see Table S-2). The 19F NMR spectra of 8-4 and 10-5 are similar in that they exhibit only two quartets, indicative of a single ribbon of seven and nine edge-sharing m- or p-C6(CF3)2 hexagons, respectively, and overall C1 symmetry. Furthermore, the 7JFF values for the quartets, 11.7 and ca. 12 Hz for 8-4 and 11.3 and 12.0 Hz for 10-5, are more similar to one another than is the case for ribbons that have a p3 end and a pmp end. The structure of 10-5 is now known to have a pmpmpmpmp addition pattern (this work), and we therefore propose that the structure of 8-4 has a pmpmpmp addition pattern. This assignment is also supported by its low relative energy (see Table S-2). X-ray Crystallography. A drawing of the structure of 10-5, with thermal ellipsoids for the CF3 groups and the cage carbon atoms to which they are attached, is shown in Figure 2. Other drawings showing thermal ellipsoids and IUPAC lowest-locant numbering for all atoms are in the SI (Figures S-6 and S-7). In harmony with the 19F NMR spectra, the CF3 groups are found along ribbons of edge-sharing p- and m-C6(CF3)2 hexagons. The addition pattern of 10-5 is a pmpmpmpmp ribbon. The X-ray structures of the other C60(CF3)n derivatives listed in Table 1 have shown that ribbons of edge-sharing p- and m-C6(CF3)2 hexagons lead to a number of very short C(sp2)-C(sp2) pentagon-hexagon junctions (PHJs) in the fullerene cage, in J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11555

Popov et al.

ARTICLES

Figure 2. Drawings of the structure of 10-5 (C1-pmpmpmpmp-C60(CF3)10 (50% probability ellipsoids for the CF3 groups and the cage carbon atoms to which they are attached). The IUPAC lowest locants are shown for some cage carbon atoms (note that this orientation is not the same as in the Schlegel diagram for this compound in some of the other figures). See Supporting Information for drawings showing the numbering of all atoms.

contrast to the long PHJs in C60, which are 1.45 ( 0.01 Å.51 There are five double bonds in pentagons (DBIPs), which are generally considered to be destabilizing:52,53 C4-C5, 1.352(3) Å; C9-C10, 1.352(3) Å; C17-C37, 1.345(3) Å; C27-C45, 1.355(3) Å; and C42-C43, 1.345(3) Å. These are of two types: (i) terminal DBIPs (t-DBIPs) that have only one C(sp2) nearest neighbor and (ii) non-terminal DBIPs (nt-DBIPs) that have a pair of C(sp2) nearest neighbors. Only the double bond is an nt-DBIP. This distinction will become important in the Discussion section. The cage C-C distances involving two sp2 C atoms in the structure of 10-5 are shown in Figure S-7 (see SI). As in the structures of other fullerene(CF3)n structures with ribbon addition patterns, there is a network of F‚‚‚F contacts between F atoms of CF3 groups that share the same hexagon in 10-5. These range from 2.522(3) to 2.678(3) Å. None of the CF3 groups in the structure of 10-5 is eclipsed with respect to the three cage C atoms that radiate from the cage C atom to which the CF3 group is attached. That observation is consistent with the lack of any 19F NMR multiplets with δ values less than 60 ppm, which is the hallmark of eclipsed CF3 groups.28,30 Electronic and Fluorescence Spectra. Peak maxima and the low-energy onsets of absorption from the electronic absorption spectra of the 18 compounds with n e 12 are listed in Table 3 along with relevant data for some other C60(X)n compounds.54-58 (51) Olmstead, M. M.; de Bettencourt-Dias, A.; Lee, H. M.; Pham, D.; Balch, A. L. Dalton Trans. 2003, 3227-3232. (52) Matsuzawa, N.; Dixon, D. A.; Fukunaga, T. J. Phys. Chem. 1992, 96, 75947604. (53) Matsuzawa, N.; Fukunaga, T.; Dixon, D. A. J. Phys. Chem. 1992, 96, 10747-10756. (54) Gonzalez, R.; Wudl, F.; Pole, D. L.; Sharma, P. K.; Warkentin, J. J. Org. Chem. 1996, 61, 5837-5839. (55) Kadish, K. M.; Gao, X.; Van Caemelbecke, E.; Suenobu, T.; Fukuzumi, S. J. Phys. Chem. A 2000, 104, 3878-3883. (56) Kitagawa, T.; Tanaka, T.; Takata, Y.; Takeuchi, K.; Komatsu, K. J. Org. Chem. 1995, 60, 1490-1491. (57) Kadish, K.; Gao, X.; Van Caemelbecke, E.; Suenobu, T.; Fukuzumi, S. J. Am. Chem. Soc. 2000, 122, 563-570. 11556 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

Absorption and emission spectra of four of the six isomers of C60(CF3)10 are shown in Figure 3 (the fluorescence spectrum of 10-1 was not recorded), and absorption spectra of the other C60(CF3)n compounds with n e 12 are shown in Figure S-8 (see SI). Electrochemical Measurements. Cyclic voltammograms for 10 of the 18 C60(CF3)n compounds with n e 12 are shown in Figures 4 and 5. Additional annotated cyclic voltammograms are shown in Figure S-9 (see SI). The E1/2 values for reversible reductions (i.e., quasi-reversible waves with ∆Ep e 90 mV for a scan speed of 20 mV s-1 and ∆Ep e 150 mV for scan speeds greater than or equal to 2.0 mV s-1) are listed in Table 4. Eleven of the 18 compounds exhibited three reversible reductions (the exceptions were 8-5, 10-2, 10-3, 10-6, 12-1, 12-2, and 12-3). The compounds 8-5 and 12-1 each exhibited only one quasireversible reduction. In a few cases (the first reductions of 8-1 and 8-5 and the second reduction of 10-3), the ratio ic/ia was less than one, but the magnitude of ic and ia as well as their ratio did not change during multiple scans. DFT-Predicted Frontier Orbitals for C60(CF3)n Compounds and C60(CF3)n- Radical Anions. The PBE-predicted relative energies of the Kohn-Sham HOMOs and LUMOs for the 22 real and four hypothetical C60(CF3)n compounds studied are listed in Table 4. Drawings of these orbitals for 10-4, 12-1, 16-2, and 18-1 are shown in Figure S-10 (see SI). The singly occupied molecular orbitals (SOMOs) of the C60(CF3)n- radical anions are topologically virtually identical to the LUMOs of the corresponding C60(CF3)n neutral compounds, as shown for 6-2, 8-1, and 10-5 in Figure S-11 (see SI). In addition, Figure 6 also includes a Schlegel representation of the SOMO for the 6-1- radical anion. The blue (+) and green (-) circles represent the upper lobes of the π atomic orbitals for each cage carbon atom scaled approximately to their contributions to the SOMO. Furthermore, the relative LUMO and SOMO energies are, as expected, strongly correlated, as shown in Figure S-12 (see SI). The DFT-predicted LUMO energies for C60(X)n species with the same addition patterns as those of 2-1, 4-1, 6-1, 8-1, and 10-2 for X ) CH3, Ph, H, NH2, CH2F, CHF2, CF3, F, NO2, and CN are shown graphically in Figure 6, and those for C60(X)n species with the same addition patterns as those of 6-2, 10-1, and 12-2 for the same ten substituents are shown in Figure S-13 (see SI). For the addition patterns in Figure 6, there is a monotonic increase in E(LUMO) from n ) 2 to n ) 10 for X ) CH3, Ph, H, NH2, and CH2F and a monotonic decrease in E(LUMO) from n ) 2 to n ) 10 for X ) CF3, F, NO2, and CN. Interestingly, the E(LUMO) values for all five addition patterns are nearly the same for X ) CHF2. As might be expected, the LUMO energies for the C60(X)n compounds are correlated with both gas-phase electron affinities (EAs) of m-X-C6H4NO2 compounds59 and group electronegativities60 of the substituents X calculated from benzene deformations, as shown in Figure S-14 (see SI) for the addition pattern of 10-2 with X ) CH3, H, NH2, CF3, F, NO2, and CN. The trends are not strictly monotonic, however, possibly because the substituents are attached to C(sp3) atoms in C60Xn derivatives (58) Coheur, P.-F.; Cornil, J.; dos Santos, D. A.; Birkett, P. R.; Lievin, J.; Bredas, J. L.; Walton, D. R. M.; Taylor, R.; Kroto, H. W.; Colin, R. J. Chem. Phys. 2000, 112, 8555-8566. (59) Kebarle, P.; Chowdhury, S. Chem. ReV. 1987, 87, 513-534. (60) Campanelli, A. R.; Domenicano, A.; Ramondo, F.; Hargittai, I. J. Phys. Chem. A 2004, 108, 4940-4948.

Investigation of Frontier Orbitals of Fullerene Derivatives

ARTICLES

Table 3. Electronic Spectral Data for C60(CF3)n (n ) 2, 4, 6, 8, 10, 12) and Related Compoundsa compd

low-energy absorption onset, nm

main absorption features, nm

2-1 1,7-C60(CO2Me)(CH2CH2SiMe3)b 1,7-C60(CH2Ph)2c 1,7-C60(tBu)(R)d 4-1 1,6,11,18-C60(CH2Ph)4e 6-1 6-2 1-Cl-6,9,12,15,18-C60Ph5 f 8-1 8-2 8-3 8-4 8-5 10-1 10-2 10-3 10-4 10-5 10-6 12-1 12-2 12-3

730

690, 626, 601, 441, 322, 256 688, 618, 538, 448, 332 ca. 690, ca. 610, ca. 540, 445, 329, 256 ca. 450, ca. 320, ca. 250 672, 610, 568, 529, 450, 350, 318, 253 ca. 690, 464, 364, 304, 250 598, 470, 370, 320 511, 463, 432, 378, 351, 335, 270, 256 516, 474, 438, 394, 351, 339, 272, 259 648, 593, 523, 477, 440, 357, 340, 330 655, 580, 544, 474, 410, 380, 313, 265, 248 652, 580, 539, 494, 472, 422, 377, 315, 280 522, 483, 349, 306 622, 540, 482, 457, 390, 362, 321, 290 665, 567, 525, 492, 459, 422, 359, 323, 298 540, 483, 456, 427, 370, 343, 326 601, 485, 454, 421, 323 580, 539, 501, 432, 375, 357, 340 535, 492, 461, 436, 383, 343/331 529, 498, 454, 350, 312, 295, 278 429, 410, 382, 338, 313, 291 637, 613, 596, 545, 470, 378, 360, 328 468, 434, 406, 328, 311

720 700 565 700 700 710 640 670 800 650 600 750 650 640 630 740 500

a All data from this work unless otherwise noted; all spectra from this work were recorded for dichloromethane solutions. b Reference 54. c Reference 55. Reference 56. e Reference 57. f Reference 58; the spectra of the 1-H- and 1-OH- analogues are very similar to the spectrum of the 1-Cl- compound listed here.

d

and C(sp2) atoms in substituted benzenes. The LUMO energies follow the order CH3 ∼ H > NH2 > CF3 > F > NO2 > CN, the m-X-C6H4NO2 EA values ((0.1 eV) follow the order NH2 < CH3 ∼ H < F < CF3 < CN < NO2,59 and the group electronegativities follow the order H < CH3 < NH2 < CF3 < CN < F < NO2.60 ESR Spectra. Experimental and DFT-predicted X-Band ESR spectra of the radical anions C1-pmp-C60(CF3)4- (4-1-) and C1p3mpmpmp-C60(CF3)10- (10-2-) are shown in Figure 7. The accompanying Schlegel diagrams depict the B3LYP-predicted SOMO for each compound. Discussion

New Addition Patterns. Of the 18 C60(CF3)n addition patterns with n e 12, only those for 2-1, 4-1, and 6-2 are known with substituents other than CF3 groups (e.g., 1,7-C60Bn2,61,62 1,7-C60(COOH)(CH2CH2SiMe3),54 1,7-C60(t-BuOO)2,63 1,6,11,18C60Bn4,57 1,6,9,12,15,18-C60Me6,64 and 1,6,9,12,15,18-C60Br665). The p3mp addition pattern of 6-1 has been observed previously, not only for C60(CF3)6 (structure determined by 19F NMR spectroscopy19 and DFT calculations20) but also for the C60like polar region in one of the isomers of C70(CF3)6 (structure determined by X-ray crystallograpy31). Trifluoromethyl groups are sterically bulky (they are significantly larger than iodine atoms, for example66,67), and of the compounds with n e 12 (61) Kadish, K. M.; Gao, X.; Van Caemelbecke, E.; Hirasaka, T.; Suenobu, T.; Fukuzumi, S. J. Phys. Chem. 1998, 100, 3898-3906. (62) Zheng, M.; Li, F.; Shi, Z.; Gao, X.; Kadish, K. J. Org. Chem. 2007, 72, 2538-2542. (63) Huang, S.; Xiao, Z.; Wang, F.; Gan, L.; Zhang, X.; Hu, X.; Zhang, S.; Lu, M.; Pan, Q.; Xu, L. J. Org. Chem. 2004, 69, 2442-2453. (64) Al-Matar, H.; Abdul-Sada, A. K.; Avent, A. G.; Fowler, P. W.; Hitchcock, P. B.; Rogers, K. M.; Taylor, R. J. Chem. Soc., Perkin Trans. 2002, 2, 53-58. (65) Troyanov, S. I.; Popov, A. A.; Denisenko, N. I.; Boltalina, O. V.; Sidorov, L. N.; Kemnitz, E. Fullerenes Nanotubes Carbon Nanostruct. 2003, 11, 61-77. (66) Charton, M. J. Am. Chem. Soc. 1969, 91, 615-618. (67) Bott, G.; Field, L. D.; Sternhell, S. J. Am. Chem. Soc. 1980, 102, 56185626.

only 6-2 and 12-3 are known to have two CF3 groups on adjacent cage carbon atoms.21 Therefore, 1,4 addition appears to be the rule for CF3 (and C2F5) groups, as it is for the addition of Br atoms.68 Most other classes of C60 derivatives made by multiple additions of a single substituent involve 1,2 additions, including hydrofullerenes,69,70 fluorofullerenes,69-71 and polycycloadducts.72 Furthermore, only two of the 18 addition patterns with n e 12 have two CF3 groups on the same pentagon (10-3 has one 1,3-C5(CF3)2 pentagon25 and 12-2 has two of them28). It is not surprising, therefore, that five of the seven new compounds reported here have one of the two more common fullerene(CF3)n addition patterns, (i) a single ribbon of edgesharing m- and/or p-C6(CF3)2 hexagons (8-1 has a p3mpmp ribbon, 8-4 has a pmpmpmp ribbon, and 10-5 has a pmpmpmpmp ribbon) or (ii) a ribbon plus an isolated p-C6(CF3)2 hexagon (8-3 and 8-5, like the known structure of 8-2,24 both have a variation of the p3mp,p addition pattern, the only difference being the location of the isolated hexagon; 10-1 has a p3mpmp,p addition pattern). The structures of 10-6 and 12-3 are unprecedented but are related to known addition-pattern motifs. Most of the C60(CF3)n addition patterns have not been the subjects of previous computational studies, even with the common substituents H, F, Cl, Br, Me, or Ph. This may be, in part, because most of the C60(CF3)n addition patterns are asymmetric, and fullerene theorists have tended to focus on C60(X)n derivatives with at least one element of symmetry. Another reason, in some cases, may be the frequent assumption that the most stable C60(X)n+2 compounds must derive from the most (68) Clare, B. W.; Kepert, D. L. J. Mol. Struct. (THEOCHEM) 1995, 340, 125142. (69) Clare, B. W.; Kepert, D. L. J. Mol. Struct. (THEOCHEM) 2002, 589590, 209-227. (70) Clare, B. W.; Kepert, D. L. J. Mol. Struct. (THEOCHEM) 2003, 621, 211231. (71) Boltalina, O. V.; Strauss, S. H. In Dekker Encyclopedia of Nanoscience and Nanotechnology; Schwarz, J. A., Contescu, C., Putyera, K., Eds.; Marcel Dekker: New York, 2004; pp 1175-1190. (72) Hirsch, A.; Brettreich, M. Fullerenes: Chemistry and Reactions; WileyVCH: Weinheim, 2005. J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11557

Popov et al.

ARTICLES

Figure 4. Cyclic voltammograms of the two isomers of C60(CF3)6 and three of the five isomers of C60(CF3)8 (0.10 M TBA+BF4- in CH2Cl2; the scan rate for the black voltammograms was 20 mV s-1; the scan rates for the red voltammograms were 5.0 V s-1 for 6-1, 3.5 V s-1 for 6-2, and 2.0 V s-1 for 8-1, 8-2, and 8-3; E1/2(C600/-) ) -0.98 or -0.46 V vs Fe(Cp)2+/0 or Fe(Cp*)2+/0 internal standards, respectively). The vertical dotted lines show the first, second, and third reduction E1/2 values for C60 under identical conditions. Slow- and fast-scan-rate cyclic voltammograms for all of the compounds investigated are collected in the Supporting Information.

Figure 3. Absorption and emission spectra of toluene solutions of five isomers of C60(CF3)10. The expansion factor is not the same for each spectrum. The insets show portions of the absorption spectra (dotted lines) and the emission spectra (solid lines). Absorption spectra of the other 13 compounds that were investigated can be found in the Supporting Information.

stable C60(X)n derivatives by simply adding two more X substituents with no rearrangement (e.g., this assumption was made in recent computational studies of C60Cln isomers73 and C60Fn isomers74). This assumption, however, has been challenged recently, both on theoretical grounds75 and in a recent study of C60F8 isomers.76 Absorption and Emission Spectra, Optical Gaps, and DFT-predicted HOMO-LUMO Gaps. The UV-vis spectra of exohedral fullerene derivatives of a fixed composition are strongly sensitive to the addition pattern of the substituents58,77-79 (73) Liang, Y.; Shang, Z.; Wang, G.; Cai, Z.; Pan, Y.; Zhao, X. J. Mol. Struct. (THEOCHEM) 2004, 677, 15-19. (74) Van Lier, G.; Cases, M.; Ewels, C. P.; Taylor, R.; Geerlings, P. J. Org. Chem. 2005, 70, 1565-1579. (75) Sandall, J. P. B.; Fowler, P. W. Org. Biomol. Chem. 2003, 1, 1061-1066. (76) Goryunkov, A. A.; et al. J. Fluorine Chem. 2006, 127, 1423-1435. (77) Foley, S.; Bosi, S.; Larroque, C.; Prato, M.; Janot, J.-M.; Seta, P. Chem. Phys. Lett. 2001, 350, 198-205. (78) Kordatos, K.; Bosi, S.; da Ros, T.; Zambon, A.; Lucchini, V.; Prato, M. J. Org. Chem. 2001, 66, 2802-2808. (79) Marchesan, S.; da Ros, T.; Prato, M. J. Org. Chem. 2005, 70, 4706-4713. 11558 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

Figure 5. Cyclic voltammograms of five of the six isomers of C60(CF3)10 (0.10 M TBA+BF4- in CH2Cl2; the scan rate was 20 mV s-1; E1/2(C600/-) ) -0.98 or -0.46 V vs Fe(Cp)2+/0 or Fe(Cp*)2+/0 internal standards, respectively). The vertical dotted lines show the first, second, and third reduction E1/2 values for C60 under identical conditions.

but, except for minor shifts in λmax values of a few nm, are not sensitive to the electronic properties of the substituents unless the substituents themselves are chromophores. This can be readily seen in Table 3 by comparing the λmax values for 2-1 with those for 1,7-C60(C6H5CH2)2 and 1,7-C60(COOCH3)(SiMe3), the λmax values for 4-1 with those for 1,6,11,18-C60Bn4, and the λmax values for 6-2 with those for 1-Cl-6,9,12,15,18C60Ph5 (the low-energy absorptions of 1,6,11,18-C60Bn4 were not reported in the original paper,57 but we recently learned that the lowest energy absorption occurs at ca. 690 nm for this compound [X. Gao, personal communication, 2007]). In addition, the ranges of DFT-predicted HOMO-LUMO gaps for the

Investigation of Frontier Orbitals of Fullerene Derivatives

ARTICLES

Table 4. Experimental and DFT Results for C60(CF3)n Derivatives (n ) 2, 4, 6, 8, 10, 12, 16, 18) relative E1/2, V vs C600/- a

DFT cmpd

C60 2-1 4-1 6-1 6-2 6-3-CF3 6-4-CF3g 8-1 8-2 8-3 8-4 8-5 8-6-CF3 10-1 10-2 10-3 10-4 10-5 10-6 12-1 12-2 12-3k 16-1 16-2 16-3 18-1 18-2-CF3

exptl λmax, eV (nm)b

E(HOMO) − E(LUMO), eV (vs C60, eV)c

E(LUMO), eV (vs C60, eV)c

0/−

−/2−

2−/3−

1.95 (637) 1.79 (690) 1.85 (672) 2.08 (598) 2.43 (511)

1.639 (0.000) 1.430 (-0.209) 1.443 (-0.196) 1.445 (-0.194) 1.859 (+0.220) 1.381 (-0.258) 1.076 (-0.563) 1.498 (-0.141) 1.482 (-0.157) 1.333 (-0.306) 1.701 (+0.062) 1.420 (+0.219) 1.302 (-0.337) 1.139 (-0.500) 1.550 (-0.089) 1.662 (-0.023) 1.636 (+0.003) 1.748 (+0.109) 1.445 (-0.194) 2.245 (+0.606) 1.614 (-0.025) 2.086 (+0.447) 1.798 (+0.159) 2.254 (+0.615) 1.897 (+0.258) 1.787 (+0.148) 2.104 (+0.465)

-4.379 (0.000) -4.592 (-0.213) -4.682 (-0.303) -4.796 (-0.417) -4.378 (+0.001) -4.858 (-0.479) -5.020 (-0.641) -4.850 (-0.471) -4.912 (-0.533) -5.017 (-0.639) -4.585 (-0.206) -4.863 (-0.486) -4.895 (-0.516) -5.129 (-0.749) -4.894 (-0.515) -4.754 (-0.375) -4.529 (-0.150) -4.638 (-0.259) -4.922 (-0.543) -4.278 (+0.101) -4.919 (-0.540) -4.331 (+0.048) -4.792 (-0.413) -4.533 (-0.154) -4.810 (-0.431) -4.848 (-0.469) -4.928 (-0.549)

0.00 {0.46} 0.11 [0.09] 0.17 [0.16] 0.26 [0.26] -0.07 [-0.08] [0.31] [0.44] 0.33 [0.30] 0.32 [0.35] 0.45 [0.44] 0.06 [0.09] 0.31 [0.31] [0.34] 0.57 [0.53] 0.32 [0.34] 0.17 [0.22] 0.07 [0.04] 0.12 [0.13] 0.33 [0.36] -0.16 [-0.21] 0.32 [0.36] -0.13 [-0.12] [0.25] [0.04] [0.27] [0.30] [0.37]

-0.40 {0.06} -0.30 [-0.32] -0.26 [-0.26] -0.28 [-0.27] -0.53 [-0.52]

-0.84 {-0.38} -0.83 -1.01d -0.93e -1.02 f

1.92 (648) 1.90 (655) 1.90 (652) 2.38 (522) 2.00 (622) 1.87 (665) 2.30 (540) 2.06 (601) 2.14 (580) 2.32 (535) 2.34 (529) 2.90 (429) 1.87 (662) 2.65 (468)

-0.19h [-0.25] -0.19 [-0.18] -0.03 [-0.06] -0.38 [-0.39] [-0.21]

ca. -0.66 h,i -0.79h -0.96h -1.06h

-0.07 [-0.03] -0.47f [-0.41] -0.44 [-0.47] -0.47 [-0.43] -0.46 [-0.48] -0.34 j [-0.30] [-0.58] -0.39 [-0.33] -0.59 [-0.55] [-0.25] [-0.43] [-0.45] [-0.23] [-0.18]

-1.10 f -1.07f -0.90f

a Reversible wave observed at 20 mV s-1 unless otherwise noted; electrolyte ) 0.10 M N(n-Bu) BF in dichloromethane; Fe(Cp) +/0 or Fe(Cp*) +/0 4 4 2 2 internal standard (E1/2 ) -0.46 or -0.98 V vs C600/-, respectively). The values shown in braces are E1/2 values vs Fe(Cp)2+/0. The values shown in italics in square brackets are (i) PBE-predicted 0/- E1/2 values vs C600/- based on a correlation with LUMO energies or (ii) B3LYP-predicted -/2- E1/2 values vs C600/- based on a correlation with the sum of second electron affinities plus ∆∆G(solvation). b Solvent ) dichloromethane. c From PBE-DFT-predicted HOMO and LUMO energies. d Scan rate ) 200 mV s-1. e Scan rate ) 5.0 V s-1. f Scan rate ) 3.5 V s-1. g The 6-4 addition pattern, 1,23,28,33,38,60C60(X)6, has only been observed for X ) CMe(COOCMe3)2 (see Figure S-3 in the Supporting Information for the Schlegel diagram of this compound). h Scan rate ) 2.0 V s-1. i The 2-/3- E j 1/2 value for 8-1 was not precisely determined because the third reduction wave was poorly defined. Scan rate ) 100 mV s-1. k This product may be a mixture of up to three isomers, all with two skew-pentagonal-pyramidal arrays of six CF3 groups on opposite C60 poles.

compounds in Figure 6 with X ) CH3, Ph, CH2F, CHF2, CF3, and CN are only 0.013, 0.032, 0.044, 0.063, and 0.137 eV for the 2-1, 4-1, 6-1, 8-1, and 10-2 addition patterns, respectively. In addition, a DFT-predicted 0.084 eV range of HOMO-LUMO gaps was recently reported34 for the set of compounds 1,23,28,33,38,60-C60(X)6 with X ) CH3, CH2F, CHF2, CF3, and CBr(COOEt)2 and CMe(COOMe)2 (the compound with X ) CMe(COOCMe3)2 is known and has been structurally characterized33). Therefore, the electronic absorption spectra reported herein may prove useful in the future for the elucidation of structures of other C60(X)n derivatives with bulky substitutents. Absorption spectra for the 18 C60(CF3)n compounds are reported here for the first time except for those of 2-1 and 6-2 (n e 12). The emission spectra of 10-2, 10-3, 10-4, and 10-5 are also reported here for the first time. The optical gaps for the C60(CF3)n compounds, which correspond to the lowestenergy absorption feature in each spectrum, do not steadily increase as n increases, unlike the situation with the series of methanofullerenes C60(CR1R2)n (n ) 1-6), for which the longest-wavelength λmax value does decrease as n increases).80 For example, the lowest-energy absorptions of 2-1, 4-1, 8-1, 8-3, and 10-1 are red-shifted relative to C60, and the analogous absorptions of 6-1, 6-2, 8-4, 10-2, 10-3, 10-4, and 10-5 are blueshifted relative to C60. More specifically, the spectra of the five (80) Cardullo, F.; Seiler, P.; Isaacs, L.; Nierengarten, J.-F.; Haldimann, R. F.; Diederich, F.; Mordasini-Denti, T.; Thiel, W.; Boudon, C.; Gisselbrecht, J. P.; Gross, M. HelV. Chim. Acta 1995, 80, 343-371.

isomers of C60(CF3)10 shown in Figure 3 are different from one another in (i) the onset of absorption and (ii) the number of detectable features. The range of optical gaps for these five isomers is 0.69 eV, the range for the five isomers of C60(CF3)8 is 0.48 eV, and the optical gaps for the two isomers of C60(CF3)6 and two isomers of C60(CF3)12 differ by 0.35 and 1.03 eV, respectively. These are extremely large optical-gap ranges for isomers of C60 derivatives with a given composition. In contrast, the lowest energy absorption is ca. 690 nm for both 1,9- and 1,7-C60(Bn)2;55 it ranges from 696 to 721 nm (∆gap ) 0.06 eV) for six regioisomers of bis(pyrrolidine) adducts of C60,81 and it ranges from 650 to 720 nm (∆gap ) 0.19 eV) for nine tris(N-methylpyrrolidine) adducts of C60.79 Even for isomers of pentakis-cycloadducts of C60 (the functional equivalent of C60(X)10 derivatives), the spectra of which were previously described as differing “distinctly from one another,” the range of optical gaps reported was only 0.15 eV.80 Nevertheless, in addition to the work reported herein, there are two previous examples for which relatively large opticalgap ranges (i.e., g 0.25 eV) for isomeric C60(X)n derivatives can be inferred. The first example is a study of hexakiscycloadduct derivatives of C60 with different addition patterns (the functional equivalent of of C60(X)12 derivatives). Although the substituents are not identical (and therefore the compounds are not strictly isomeric), the lowest energy absorption varied (81) Kordatos, K.; da Ros, T.; Prato, M.; Bensasson, R. V.; Leach, S. Chem. Phys. 2003, 293, 263-280. J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11559

Popov et al.

ARTICLES

Figure 6. DFT-predicted LUMO energies for single-substituent C60(X)n derivatives with the same addition patterns as 2-1, 4-1, 6-1, 8-1, and 10-2 and with a variety of X substituents. The Schlegel diagram in the lowerleft corner shows the five addition patterns used for each substituent. The addition pattern for 10-2 is all ten black circles; the addition pattern for 8-1 is all black circles except for the two labeled “10”; etc. The Schlegel diagram in the upper-right corner shows the singly occupied molecular orbital (SOMO) for the 6-1- radical anion with CF3 groups. The blue (+) and green (-) circles represent the upper lobes of the π atomic orbitals for each cage carbon atom scaled approximately to their contributions to this orbital. The SOMO for the 8-1- radical anion with CF3 groups is nearly identical to the 6-1- SOMO.

from 600 to 450 nm (∆gap ) 0.69 eV).82 The second example is the emerald-green compound 1,23,28,33,38,60-C60(CMe(COOCMe3)2)6, which has an intense band at 850 nm and a lowest energy absorption at 1170 nm.33 If the substituent effect on the optical gap is ignored, the difference between the lowest energy absorption of this compound and the 516-nm onset of absorption of 1-Cl-6,9,12,15,18-C60Ph558 results in a ∆gap value of 1.34 eV. Finally, as part of this work, we recorded the absorption spectra of the two isomers of C60F8,76 and these are shown in Figure S-15. The lowest energy absorption are ca. 650 nm and ca. 575 nm, resulting in a ∆gap value of 0.25 eV. To confirm that the lowest-energy feature in the absorption spectrum (i.e., the optical gap) of a given C60(CF3)n derivative may, in general, be unambiguously assigned to the HOMOLUMO transition,83-89 we measured the emission spectra of four (82) Bourgeois, J.-P.; Woods, C. R.; Cardullo, F.; Habicher, T.; Nierengarten, J.-F.; Gehrig, R.; Diederich, F. HelV. Chim. Acta 2001, 84, 1207-1226. (83) Coheur, P.-F.; Cornil, J.; dos Santos, D. A.; Birkett, P. R.; Lievin, J.; Bredas, J. L.; Walton, D. R. M.; Taylor, R.; Kroto, H. W.; Colin, R. J. Chem. Phys. 2000, 112, 6371-6381. (84) Guldi, D.; Asmus, K. D. J. Phys. Chem. A 1997, 101, 1472-1481. (85) Lebedkin, S. F.; Rietschel, H.; Adams, G. B.; Page, J. B.; Hull, W. E.; Hennrich, F. R.; Eisler, H.-J.; Kappes, M. M.; Kratschmer, W. J. Chem. Phys. 1999, 110, 11768-11778. (86) Amarantov, S. V.; Bezmelnitsin, V. N.; Boltalina, O. V.; Danailov, M.; Dudin, P. V.; Ryzkov, A. V.; Stankevich, V. G. Nucl. Instrum. Methods Phys. Res., Sect. A 2001, 470, 318-322. 11560 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

Figure 7. Experimental (dichloromethane, 25 °C) and DFT-predicted X-Band ESR spectra of the radical anions C1-pmp-C60(CF3)4- (4-1-; dichloromethane) and C1-p3mpmpmp-C60(CF3)10- (10-2-; o-dichlorobenzene). The DFT-computed hyperfine coupling constants were uniformly scaled by 0.9 in order to match the experimental values. The Schlegel diagrams show the B3LYP-predicted singly occupied molecular orbital (SOMO) for each compound. The blue (+) and green (-) circles represent the upper lobes of the π atomic orbitals for each cage carbon atom scaled approximately to their contributions to the SOMO.

isomers of C60(CF3)10 and compared them with the corresponding absorption spectra (see Figure 3). The emission spectra were independent of the excitation wavelength used, indicating that emission occurs from the S1 state. It should be noted that the “mirror-image rule” is not always obeyed by fullerene derivatives because of the high density of excited states and the concomitant overlap of multiple absorption bands.58 Nevertheless, there is a good correlation between the absorption and emission spectra for 10-2, 10-3, and 10-4, confirming the assignment of the HOMO-LUMO transition in these derivatives. The broad featureless band at 567 nm in the emission spectrum of 10-2 can be matched with the analogous broad band in the absorption spectrum at 550 nm. The emission spectra of 10-3 and 10-4 have more structure and are therefore even more convincing. For 10-3, the emission bands at 615 and 673 nm (87) Green, J., W. H.; Gorun, S. M.; Fitzgerald, G.; Fowler, P. W.; Ceulemans, A.; Titeca, B. C. J. Phys. Chem. 1996, 100, 14892-14898. (88) Cramariuc, O.; Hukka, T. I.; Rantala, T. T.; Lemmetyinen, H. J. Phys. Chem. A 2006, 110, 12470-12476. (89) Guldi, D. M.; Prato, M. Acc. Chem. Res. 2000, 33, 695-703.

Investigation of Frontier Orbitals of Fullerene Derivatives

Figure 8. Plot of DFT-predicted Kohn-Sham HOMO and LUMO energies for C60 and C60(CF3)n derivatives. The horizontal lines are the DFT-predicted HOMO and LUMO energies for C60. The compound 18-2 is not known with CF3 groups at this time and therefore represents a hypothetical derivative.

(∆E ) 0.175 eV) correspond to the absorption bands at 601 and 554 nm, respectively (∆E ) 0.175 eV); for 10-4, the emission bands at 613, 664 and 727 nm (∆E(613/664) ) 0.156 eV; ∆E(664/727) ) 0.162 eV) correspond to the absorption bands at 584, 540, and 503 nm, respectively (∆E(584/540) ) 0.173 eV; ∆E(540/503) ) 0.169 eV). In the case of 10-5, the main emission band at 563 nm confirms that the 535-nm band in the absorption spectrum, which is better resolved in dichloromethane (not shown) than in toluene, can be assigned to the HOMO-LUMO transition. Figure S-16 shows a correlation of DFT-predicted HOMOLUMO gaps with optical gaps for the 18 C60(CF3)n derivatives dissolved in dichloromethane. All of the optical gaps are in a region with unit slope and a width of 0.4 eV on the HOMOLUMO gap axis. The overall correlation between the experimental and calculated frontier-orbital energy gaps is obvious. Deviations of (0.2 eV from a line of unit slope bisecting the region are not surprising since the HOMO-LUMO gap does not take into account geometric changes of the molecule in the excited state, the possible admixture of other configurations, or solvent effects.90 Figure 8 displays the HOMO and LUMO energies for C60 and 23 of the compounds listed in Table 4 on the same energy axis. The E(HOMO) values become more negative as n increases: the average E(HOMO) values for the compositions with n ) 6, 8, 10, 12, and 16 (all real compounds) show a steady decrease except that the average E(HOMO) values for n ) 8 and 10 are nearly the same. On the other hand, the E(LUMO) values vary in an irregular way. There is a much larger difference in the range of E(LUMO) values for isomers of a given composition than in the range of E(HOMO) values. These ranges are: for n ) 6, ∆E(HOMO) ) 0.00 eV, ∆E(LUMO) ) 0.42 eV; for n ) 8, ∆E(HOMO) ) 0.12 eV, ∆E(LUMO) ) 0.43 eV; for n ) 10, ∆E(HOMO) ) 0.28 eV, ∆E(LUMO) ) 0.60 eV; for n ) 12, ∆E(HOMO) ) 0.01 eV, ∆E(LUMO) ) 0.64 eV; and for n ) 16, ∆E(HOMO) ) 0.19 eV, ∆E(LUMO) ) 0.28 eV (however, the 0.32 eV difference in E(HOMO) values for the real compound 18-1 and the hypothetical compound 18-2 is actually larger than the 0.08 eV difference in their E(LUMO) values). In addition, DFT calculations for the 24 IPR isomers (90) Armstrong, R. S.; Gallagher, R. T.; Noviandri, I.; Lay, P. A. Fullerene Sci. Technol. 1999, 7, 1003-1028.

ARTICLES

of C84 revealed a range of E(HOMO) values, 0.60 eV, only two-thirds as large as the 0.96 eV range of E(LUMO) values. Whether this is a general phenomenon for most fullerene(X)n isomers is not clear at this time. However, it appears to be general for C60(X)n compounds with the 2-1, 4-1, 6-1, 8-1, or 10-2 addition patterns for all the substituents X shown in Figure 6. Although the E(LUMO) values (and the E(HOMO) values) vary by up to 3 eV as the substituent is changed, the HOMOLUMO gaps only varied by 0.05 eV. Electrochemical Potentials and DFT-Predicted E(LUMO) Values. I. General Comments. All 18 C60(CF3)n compounds studied by cyclic voltammetry exhibited reversible first reductions (as defined earlier) at a scan speed of 20 mV s-1 in dichloromethane at 25 °C. This is in contrast to many C60Fn compounds, which generally require much higher scan speeds to achieve reversibility or do not exhibit reversible reductions at any scan speed (one exception is C60F48, which exhibits a reversible first reduction at 200 mV s-1 in dichloromethane at 25 °C).15,91-94 For reasons that are not clear at this time, some of the C60(CF3)n compounds did not exhibit reversible second or third reductions. However, it is noteworthy that 6-2 exhibited three reversible reductions. The compound 1-Cl-6,9,12,15,18-C60Ph5 (Cs-C60Ph5Cl), with the 6-2 addition pattern, has also been studied electrochemically.95 Unlike 6-2, C60Ph5Cl exhibits an irreversible two-electron reduction, even at -80 °C in dichloromethane, with an cathodic peak potential of -0.09 V vs C600/-.95 This has been attributed to the following reaction:

1-Cl-6,9,12,15,18-C60Ph5Cl + 2e- f 6,9,12,15,18-C60Ph5-+ ClThe cyclopentadienyl-like anion C5V-C60Ph5- is an especially stable derivative96,97 and is known to exhibit a reversible oneelectron reduction as well as a reversible one-electron oxidation.98 Interestingly, the compound Cs-C60Ph5H, which also undergoes irreversible reduction at 20 °C, exhibits two reversible reductions at -78 °C in THF.98 The reversible reductions of 6-2 are probably due to the stronger bonds between the cage and CF3 groups than between the cage and H or Cl atoms (this may also account for the exceptional thermal stabilities of fullerene(CF3)n compounds). In the remainder of this paper, the discussion of C60(CF3)n electrochemical potentials and their comparison to literature values of other C60(X)n will be limited to reVersible or quasireVersible potentials (i.e., E1/2 values). Since we are interested in comparing the experimental potentials with calculated LUMO energies, we will omit irreversible reductions in this analysis. The electrochemical literature on fullerenes and exohedral fullerene(X)n derivatives is extensive (and many papers also (91) Liu, N.; Touhara, H.; Morio, Y.; Komichi, D.; Okino, F.; Kawasaki, S. J. Electrochem. Soc. 1996, 143, L214-L217. (92) Liu, N.; Morio, Y.; Okino, F.; Touhara, H.; Boltalina, O. V.; Pavlovich, V. K. Synth. Met. 1997, 86, 2289-2290. (93) Zhou, F.; Van Berkel, G. J.; Donovan, B. T. J. Am. Chem. Soc. 1994, 116, 5485-5486. (94) Paolucci, D.; Paolucci, F.; Marcaccio, M.; Carano, M.; Taylor, R. Chem. Phys. Lett. 2004, 400, 389-393. (95) Birkett, P. R.; Taylor, R.; Wachter, N.; Carano, M.; Paolucci, F.; Roffia, S.; Zerbetto, F. J. Am. Chem. Soc 2000, 122, 4209-4212. (96) Matsuo, Y.; Tahara, K.; Nakamura, E. Chem. Lett. 2005, 34, 1078-1079. (97) Sawamura, M.; Nagahama, N.; Toganoh, M.; Hackler, U. E.; Isobe, H.; Nakamura, E.; Zhou, S.-Z.; Chu, B. Chem. Lett. 2000, 29, 1098-1099. (98) Iikura, H.; Mori, S.; Sawamura, M.; Nakamura, E. J. Org. Chem. 1997, 62, 7912-7913. J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11561

Popov et al.

ARTICLES

include computational studies at various levels of theory).8-13,99 Some studies compared E1/2 values (i) for different fullerene cages (e.g., C70 vs C76-D2 vs isomers of C78100), (ii) for different substituents with a fixed C60+y(X)n n value and addition pattern (e.g., different monomethano101,102 or other monoadduct103 derivatives of C60), and (iii) for different values of n with the same substituent, but with only one addition pattern for each value of n (e.g., C70Ph2,4,6,8,10104). Prior to the publication of these C60(CF3)n results (some of which were briefly communicated in 200515), there were only a few studies in which two or more E1/2 values were compared for different isomers of a giVen C60(X)n composition, and these are listed in Table 5,55,57,62,94,105-113 which also includes the ranges for C60(CF3)6,8,10 isomers from this study. It is clear from the data in Table 5 that well-characterized isomeric C60(X)n compounds can have ranges of 0/-, 1-/2-, and 2-/3- electrochemical potentials that are significantly larger than previously observed. The C60(CF3)n results have set a new standard for comparing the electrochemical potentials of isomeric fullerene compounds. For example, the 0/- E1/2 range of 0.07 V for three isomers of C78(C(COOEt)2)3 previously led to the conclusion that one of the isomers “is much more difficult to reduce than the other two derivatives” [emphasis added].100 The “tunability” of fullerene redox potentials, which is often one of the justifications for the synthesis of new fullerene derivatives,8,10,11,14,15,62,92,94,99-105,107,114-117 can be as large as 0.5 V for different isomeric exohedral fullerene derivatives of two different compositions (n ) 10 and 12). This unanticipated result demonstrates the range of electronic compliance of a π system with 20 or more conjugated double bonds on a curved 3-connected network of carbon atoms. II. Correlation of Electrochemical Potentials with DFTPredicted LUMO Energies. A plot showing the linear correlation of 0/- E1/2 values for the 18 C60(CF3)n compounds with (99) Yanilkin, V. V.; Gubskaya, V. P.; Morozov, V. I.; Nastapova, N. V.; Zverev, V. V.; Berdnikov, E. A.; Nuretdinov, I. A. Russ. J. Electrochem. 2003, 39, 1147-1165. (100) Boudon, C.; Gisselbrecht, J. P.; Gross, M.; Herrmann, A.; Ruttimann, M.; Crassous, J.; Cardullo, F.; Echegoyen, L.; Diederich, F. J. Am. Chem. Soc. 1998, 120, 7860-7868. (101) Keshavzarz-K, M.; Knight, B.; Haddon, R. C.; Wudl, F. Tetrahedron 1996, 52, 5149-5159. (102) Ocafrain, M.; Herranz, M. A.; Marx, L.; Thilgen, C.; Diederich, F.; Echegoyen, L. Chem. Eur. J. 2003, 9, 4811-4819. (103) Irngartinger, H.; Fettel, P. W.; Escher, T.; Tinnefeld, P.; Nord, S.; Sauer, M. Eur. J. Org. Chem. 2000, 455-465. (104) Avent, A. G.; Birkett, P. R.; Carano, M.; Darwish, A. D.; Kroto, H. W.; Lopez, J. O.; Paolucci, D.; Roffia, S.; Taylor, R.; Wachter, N.; Walton, D. R. M.; Zerbetto, F. J. Chem. Soc., Perkin Trans. 2 2001, 140-145. (105) Carano, M.; da Ros, T.; Fanti, M.; Kordatos, K.; Marcaccio, M.; Paolucci, D.; Prato, M.; Roffia, S.; Zerbetto, F. J. Am. Chem. Soc. 2003, 125, 71397144. (106) Caron, C.; Subramanian, R.; D’Souza, F.; Kim, J.; Kutner, W.; Jones, M. T.; Kadish, K. J. Am. Chem. Soc. 1993, 115, 8505-8506. (107) Echegoyen, L. E.; Djojo, F. D.; Hirsch, A.; Echegoyen, L. J. Org. Chem. 2000, 65, 4994-5000. (108) Boulas, P.; D’Souza, F.; Henderson, C. C.; Cahill, P. A.; Jones, M. T.; Kadish, K. J. Phys. Chem. 1993, 97, 13435-13437. (109) Azamar-Barrios, J. A.; Dennis, T. J. S.; Sadhukan, S.; Shinohara, H.; Scuseria, G. E.; Penicaud J. Phys. Chem. A 2001, 105, 4627-4632. (110) Hall, M. H.; Shevlin, P.; Lu, H.; Gichuhi, A.; Shannon, C. J. Org. Chem. 2006, 71, 3357-3363. (111) Pauloucci, F.; Marcaccio, M.; Roffia, S.; Orlandi, G.; Zerbetto, F.; Prato, M.; Maggini, M.; Scorrano, G. J. Am. Chem. Soc. 1995, 117, 65726580. (112) Meier, M. S.; Poplawska, M.; Compton, A. L.; Shaw, J. P.; Selegue, J. P.; Guarr, T. F. J. Am. Chem. Soc. 1994, 116, 7044-7048. (113) Kadish, K.; Gao, X.; Gorelik, O.; Van Caemelbecke, E.; Suenobu, T.; Fukuzumi, S. J. Phys. Chem. A 2000, 104, 2902-2907. (114) Fan, L.; Li, Y.; Li, F.; Li, Y.; Zhu, D. Chem. Phys. Lett. 1998, 294, 443446. (115) Claves, D.; Giraudet, J.; Hamwi, A.; Benoit, R. J. Phys. Chem. B 2001, 105, 1739-1742. 11562 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

Table 5. Experimental E1/2 Values for Isomeric C60+y(X)n Derivativesa C60+y(X)n

no. of isomers

0/− E1/2 range, V

C84b C78c,d C78(C(COOEt)2)3c,e C70(COOCH2COOMe)2c C70H2g C70Bn2h C70(c-ONCR)i C60(c-CH2N(TEG)CH2)2 j C60F36k C60Me2l C60(CH2Ph)2m C60(CH2Ph)4n C60(C(CO2Et)2)3o 1,2- vs 1,9-C60(CRR′)p

6 2 3 3 2 3 3 8 3 2 2 2 7 4 pairs

0.36 0.06 0.07 0.04 0.01 0.15 ca. 0 0.14 0.08 0.00 0.01 0.05 0.16 0.11, 0.03, 0.14, 0.09 0.01 0.33 0.39 0.50 0.49

(CRR′)s

1,2- vs 1,9-C60 C60(CF3)6 C60(CF3)8 C60(CF3)10 C60(CF3)12

2 2 5 6 2

−/2− E1/2 range, V 0.33 0.10 0.06 0.02 0.04 0.10 ca. 0 0.15 0.00 0.05 0.03 0.08 [0.20], [0.22], [0.18], [0.23]q 0.01 0.25 0.35 0.40 t

2−/3− E1/2 range, V 0.31 0.02 -f 0.05 0.03 0.00 ca. 0 0.11 0.00 0.05 0.03 [0.19], [0.19], [0.16], [0.25]r 0.05 0.09 0.40 u 0.20 u

a The C (CF ) data are from this work; only ranges of reversible 60 3 n potentials are listed; the literature E1/2 values were generally listed as (0.01 b V. Six cage isomers; ref 109. c Reference 100. d C2V and D3 cage isomers. e Different cage and addition-pattern isomers. f This range, reported to be 0.34 V, is complicated by the facts that the third reduction for one isomer was not reversible and was badly resolved for another isomer. g Reference 113. h 3,4 and 15,16 isomers; ref 108. i Reference 112. j Reference 105.k C1, C3, and T isomers; ref 94. l Reference 106. m 1,9 and 1,7 isomers; ref 62. n 1,6,11,18 and 1,6,9,18 isomers; ref 57. o Reference107. p Reference 110. q These E r 1/2 ranges are for the 1+/0 couples. These E1/2 ranges are for the 2+/1+ couples. s Reference 111. t This range is for only five of the isomers. u This range is for only four of the isomers.

n e 12 with the corresponding PBE-predicted LUMO energies (E(LUMO)) values is shown in Figure 9 (R2 ) 0.98; an equally good correlation (not shown; R2 ) 0.99) was found for a plot of the same E1/2 values vs the B3LYP-predicted LUMO energies). The E1/2 and E(LUMO) values in Figure 9 vary from -0.16 to 0.57 V vs C600/- (∆ ) 0.73 V) and from 0.101 to -0.749 eV vs C60 (∆ ) 0.850 eV), respectively. Linear 0/E1/2 vs E(LUMO) plots have been previously reported in at least three electrochemical studies of fullerene cycloadducts.14,80,118 In two of these studies, the range of E1/2 values was less than or equal to 0.20 V.14,118 In the third study, the range of E1/2 values was ca. 0.8 V, and the E(LUMO) values were determined for simpler model compounds with H atoms and CH2 groups representing the actual cycloaddition moieties.80 A fourth paper described an electrochemical study of higher fullerenes and included a plot of [E(HOMO) - E(LUMO)] vs [E1/2(ox) E1/2(red)] values, but a simple plot of 0/- E1/2 vs E(LUMO) values was not shown.119 With E1/2 values for multiple isomers of C60(CF3)6, C60(CF3)8, C60(CF3)10, and C60(CF3)12, it is now clear that, in general, electrochemical potentials are not a monotonic function of n, the number of CF3 radicals that were added to C60. A plot of the 18 experimental C60(CF3)n 0/- E1/2 values vs n is shown in the inset in Figure 9. Also included in the inset are 0/- E1/2 values for 4-2-CF3 and 6-3-CF3, calculated using the linear least-squares equation shown in the 0/- E1/2 vs E(LUMO) plot (116) Illescas, B. M.; Martin, N. J. Org. Chem. 2000, 65, 5986-5995. (117) Deng, F.; Wang, G.-W.; Zhang, T.-H.; Jiao, L.-J.; Chen, S. Chem. Commun. 2004, 1118-1119. (118) Zhou, J.; Rieker, A.; Grosser, T.; Skiebe, A.; Hirsch, A. J. Chem. Soc., Perkin Trans. 1997, 2, 1-5. (119) Yang, Y.; Arias, F.; Echegoyen, L.; Chibante, L. P. F.; Flanagan, S.; Robertson, A.; Wilson, L. J. J. Am. Chem. Soc. 1995, 117, 7801-7804.

Investigation of Frontier Orbitals of Fullerene Derivatives

ARTICLES

in Figure 9. (The p3-C60(X)4 addition pattern of 4-2-CF3 has been observed for X ) 9-fluorenyl;120 the as-yet unknown compound 4-2-CF3 is predicted to be 8.2 kJ mol-1 more stable than 4-1.19) The justification for including these hypothetical compounds in our analysis is the very good linear correlation of the experimental 0/- E1/2 values with the PBE-predicted E(LUMO) values. Note that 6-1 and 6-2, which have 0/- E1/2 values that differ by 0.33 V, can be envisioned as arising from the hypothetical common intermediate 4-2-CF3 by adding two CF3 groups to different pairs of cage carbon atoms (such comparisons are only meant to highlight structural relationships; nothing is implied here about the possible mechanism(s) by which 6-1 and 6-2 were formed). Similarly, 10-1, 10-2, and 10-3, which have 0/E1/2 values of 0.57, 0.32, and 0.17 V vs C600/-, respectively, can arise from 8-1 by adding two CF3 groups to different pairs of cage carbon atoms. In fact, 10-1 and 10-3 differ in the placement of only one CF3 group (see Figure 1), yet their

reduction potentials differ by 0.40 V. Furthermore, the hypothetical transformation 8-1 + 2 CF3 f 10-1 results in an increase in 0/- E1/2, while the transformation 8-1 + 2 CF3 f 10-3 results in a decrease in 0/- E1/2. Interestingly, the compound 6-2, which has six electronwithdrawing CF3 substituents, is harder to reduce than C60, and 12-1 and 12-3, with twice as many CF3 groups as 6-2, are 0.16 and 0.13 V harder to reduce than C60, respectively. In contrast, the compound 12-2 is 0.32 V easier to reduce than C60. It appears that the addition pattern is at least as important as, if not more important than, the value of n in determining the electron-acceptor properties of a particular C60(X)n compound. There are more isomers of C60(CF3)12 than the three that have been characterized to date (i.e., 12-1, 12-2, and 12-3),28 and the same may be true for C60(CF3)16 and C60(CF3)18.32 Nevertheless, none of the calculated 0/- E1/2 values listed in Table 4 for isomers with n ) 16 and 18 is more positive than the 0/E1/2 value of 10-1 or even 8-3. It may be that saturation of the fullerene π system for C60(CF3)n derivatives with n > 10 places a limitation on how positive the 0/- E1/2 value can be. Alternatively, with more than 12 CF3 groups, addition-pattern fragments that lead to low LUMO energies may be uncommon. This issue will be considered after a discussion of substituent effects. III. Correlation of LUMO Energies with Substituent Electronic Properties. Figure 6 shows the E(LUMO) values for 50 C60(X)n compounds, and Figure S-14 shows the E(LUMO) values for 29 others. Except for the X ) CF3 derivatives and for C60(CH3)6 with the 6-2 addition pattern,64 the remaining compounds are hypothetical derivatives. The main significance of these results is not that they can be used to estimate E1/2 values for nonexistent fullerene derivatives, it is because the observed trends led us to perform additional calculations and to discover a potentially general rule about the electron acceptor properties of exohedral fullerene derivatives. The remaining π system in a fullerene(X)n compound is insulated from the electronic effects of the substituents by an sp3 cage carbon atom. Therefore, for a given addition pattern, the LUMO energy should depend on the σ electron-donating or electron-withdrawing property of X (as well as on the cumulative saturation of the fullerene π system, which is probably independent of the nature of X). Even though the electronegativity of the cyano group is lower than those of the nitro group and a fluorine atom,60 cyano groups attached to sp3 cage carbon atoms121 or methano carbon atoms101 are known to produce strong anodic shifts in the first reduction relative to C60. It was concluded that “the cyano substituent appears to be more electron-withdrawing than predicted by the Hammett [σm] relation.”101 The methyl derivatives have the least negative E(LUMO) values among the ten substituents studied for the eight addition patterns shown in Figures 6 and S-14, and the cyano derivatives have, with one minor exception, the most negative E(LUMO) values (the only exception is the 6-2 addition pattern, for which the 6-2-NO2 E(LUMO) value is 0.032 eV more negative than the 6-2-CN E(LUMO) value). It is immediately apparent from the addition-pattern plots in Figure 6 that the difference between E(LUMO(C60(CH3)n)) and E(LUMO(C60(CN)n)) increases as n increases from 2 to 10. This

(120) Murata, Y.; Shiro, M.; Komatsu, K. J. Am. Chem. Soc. 1997, 119, 81178118.

(121) Keshavzarz-K, M.; Knight, B.; Srdanov, G.; Wudl, F. J. Am. Chem. Soc. 1995, 117, 11371-11372.

Figure 9. Plot of E1/2 value for the C60(CF3)n0/- couple (0.10 M TBA+BF4in CH2Cl2; E1/2(C600/-) ) -0.98 or -0.46 V vs Fe(Cp)2+/0 or Fe(Cp*)2+/0 internal standards, respectively) vs the DFT-predicted LUMO energy. The linear least-squares fit to the data is shown at the bottom of the plot. The inset is a plot of the same E1/2 values vs n, the number of CF3 groups. The black squares represent E1/2 values calculated using the least-squares equation in the main plot. Each arrow connects two compounds that are related by the addition of two (or four) CF3 groups to specific positions. The arrows are intended to show structural relationships (cf. Schlegel diagrams in Figure 1) and not to imply that the mechanism of formation of the indicated compounds is by simple addition to two CF3 groups to a fixed precursor (although this may, in the future, be shown to be true in some cases).

J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11563

Popov et al.

ARTICLES

Figure 10. Plots of differences in E(LUMO) for pairs of C60(CH3)n and C60(CN)n compounds with identical addition patterns (red circles) and for pairs of C60(CF3)n and C60(CN)n compounds with identical addition patterns (blue circles) versus n, the number of substituents. There are two red and blue points for n ) 2 and 12, three red and blue points for n ) 4, four red and blue points for n ) 6, five red and blue points for n ) 8 and 10 (the compound 10-6 was omitted), two red and blue points for n ) 12 (12-3 was omitted), and one point for n ) 24. The Schlegel diagram for Th-C60(X)24 (with the 24-1-Br addition pattern) and partial Schlegel diagrams for 1,9-C60(X)2 and 1,2,9,12-C60(X)4 (the 2-2-CN and 4-3-F addition patterns, respectively) are shown as insets.

is a sensible result: more substituents have a larger effect on the electronic structure of the fullerene derivative. It is also apparent that the increase in E(LUMO) from C60(CH3)n to C60(CH3)n+2 for these particular addition patterns is nearly constant for n ) 2, 4, 6, and 8 (0.23 ( 0.03 eV) and is equal in magnitude to the nearly constant decrease in E(LUMO) from C60(CN)n to C60(CN)n+2 (0.24 ( 0.03 eV) for the same addition patterns. However, this was not found to be generally true: the difference between the E(LUMO) values for 6-2-CH3 and 102-CH3, which is 0.29 eV, is much smaller than the corresponding difference for 6-2-CN and 10-2-CN, which is 0.84 eV. By far the most intriguing result of our DFT E(LUMO) investigation was that the difference between E(LUMO(C60(CH3)n)) and E(LUMO(C60(CN)n)) was virtually the same for the 10-1 and 10-2 addition patterns (2.49 and 2.48 eV, respectively) and varied by only 0.24 eV for the 6-1 and 6-2 addition patterns (1.59 and 1.35 eV, respectively). We subsequently calculated E(LUMO(C60(CH3)n)) and E(LUMO(C60(CN)n)) values for all of the addition patterns listed in Table 1 as well as for two addition patterns that have two or four CH3 and CN substituents on contiguous cage carbon atoms (these addition patterns are known for C60H2, C60F2, C60(CN)2, C60(Bn)2, and C60H4 and C60F4). The upper plot in Figure 10 (which includes Schlegel diagrams for the 2-2-CN,121 4-3-F,122 and 241-Br65,123 addition patterns) shows that the difference between (122) Boltalina, O. V.; Darwish, A. D.; Street, J. M.; Taylor, R.; Wei, X.-W. Perkin Trans. 2002, 2, 251-256. (123) Tebbe, F. N.; Harlow, R. L.; Chase, D. B.; Thorn, D. L.; Campbell, G. C.; Calabrese, J. C.; Herron, N.; Young, R. J.; Wasserman, E. Science 1992, 256, 822-825. 11564 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

E(LUMO(C60(CH3)n)) and E(LUMO(C60(CN)n) is remarkably constant for all addition patterns (i.e., all isomers) with a given value of n. For example, the difference [E(LUMO(C60(CH3)n)) - E(LUMO(C60(CN)n))] is 1.47 ( 0.12 eV for the four n ) 6 isomers, 1.99 ( 0.08 eV for the five n ) 8 isomers, 2.47 ( 0.09 eV for five n ) 10 isomers (10-6 was excluded), and 2.78 ( 0.01 eV for two n ) 12 isomers (12-3 was excluded). The E(LUMO) differences between the X ) CH3 and X ) CF3 compounds (not shown) and between the X ) CF3 and X ) CN compounds (also shown in Figure 10) have similarly small ranges for isomers with a given value of n, demonstrating that the upper plot in Figure 10, with regular, slightly decreasing, intervals between neighboring values of n up to n ) 24, is not the accidental result of arbitrarily choosing X ) CH3 and CN as the σ-effect extremes; this appears to be a general rule for C60(X)n derivatives, at least for the particular addition patterns we have investigated so far. Whether the rule is valid for all realistic addition patterns (i.e., those with at least one isolated example for any X substituent), whether there is a corresponding rule for isomers of cycloadducts with the same substituent and n value, and whether similar rules are found for exohedral derivatives of fullerenes other than C60 or for endohedral fullerenes remain to be seen. The implications of the electrochemical and DFT results so far are that (i) the E1/2 values are not a simple function of n but instead depend critically on the particular addition pattern (to the point where a C60(X)n compound with a given X and n value can be either easier or harder to reduce than C60, depending on its addition pattern) and (ii) the range of possible E1/2 values for C60(X)n compounds with an assortment of substituents X depends on the value of n in a uniform manner and is nearly independent of the addition pattern. For a given set of substituents, and to a good degree of approximation, one can imagine a one-dimensional E1/2 scale with variable ∆E1/2 ranges that depend on n but not on the addition pattern, anchored at particular E1/2 values that depend on the addition pattern but not on n. IV. Correlation of LUMO Energies, Addition Patterns, and Double Bonds in Pentagons. The DFT code with the PBE functional used extensively in this study for geometry optimization significantly shortens the optimization time for molecules as large as C60(CF3)n relative to the more commonly used B3LYP functional. We have already shown that PBE-predicted E(LUMO) values correlate very well with 0/- E1/2 values. Since so much of the discussion that follows depends on the interpretation of PBE-predicted cage C-C bond distances (for those compounds that have not had their structures determined by X-ray crystallography), further validation of the PBE functional is warranted. Figure S-17 shows plots of X-ray vs PBE-predicted cage C-C distances for 10-5 (this work), 102,26 and C74(CF3)12.124 The agreement is excellent; fewer than five of the 90 X-ray cage C-C distances for each compound deviate from the PBE-predicted distances by more than (3σ. Figure S-17 also shows a plot of PBE- vs B3LYP-predicted cage C-C distances for C74(CF3)12, none of which differs by more than 0.003 Å.124 Furthermore, the conformations of the six pairs of symmetry-related CF3 groups in the X-ray structure (124) Shustova, N. B.; Popov, A. A.; Newell, B. S.; Miller, S. M.; Anderson, O. P.; Seppelt, K.; Bolskar, R. D.; Boltalina, O. V.; Strauss, S. H. Angew. Chem., Int. Ed. 2007, 46, 4111-4114.

Investigation of Frontier Orbitals of Fullerene Derivatives

of C74(CF3)12, defined as F-C-C-C torsion angles, were matched by the PBE-predicted F-C-C-C torsion angles to within 1° for four pairs and to within 5° (54.4(1)° vs 51.5°) and 7° (47.1(1)° vs 54.0°) for the other two pairs. Finally, the shapes of the LUMOs or SOMOs discussed below (i.e., the relative cage-carbon atomic orbital contributions to these molecular orbitals) were plotted using the B3LYP functional instead of the PBE functional only because B3LYP gave unpaired spin densities that better matched the experimental ESR spectra of 4-1- and 10-2- (see Figure 7). (Recall that the LUMO and SOMO for a given compound are virtually indistinguishable (see Figure S-12 for three examples); we also observed that plots of the PBE- and B3LYP-predicted LUMOs were visually indistinguishable). In summary, one can be confident that the calculated cage C-C distances discussed below and cage-carbon atomic orbital contributions shown in the figures correspond very closely to the actual distances and unpaired spin densities on individual carbon atoms in C60(X)n molecules and radical anions. The 0/- E1/2 potentials in Table 4 (or their corresponding E(LUMO) values) and the Schlegel diagrams in Figure 1 are correlated in ways that appear complicated at first but become sensible after closer inspection. Consider first the isomers with n ) 8 and 10. These are 11 compounds with similar numbers of CF3 groups, and hence a similar collective electronwithdrawing inductive effect, with first-reduction potentials that range from 0.06 to 0.57 V vs C600/-. The 11 potentials can be divided into four ranges: 0.09 ( 0.03 V (8-4, 10-4, and 10-5); 0.17 V (10-3); 0.32 ( 0.01 V (8-1, 8-2, 8-5, 10-2, and 10-6); and g0.45 V (8-3 and 10-1). All seven compounds with potentials greater than 0.30 V have two conjugated nt-DBIPs. In every case, these nt-DBIPs are also in conjugation with a common hex-hex junction double bond, forming a fulvenelike fragment, and for all seVen compounds the fulVene fragment essentially defines the LUMO, as shown for 6-1 (the LUMO is virtually identical to the SOMO for the 6-1- radical anion), for 4-1 and 10-2 in Figure 7 (see also Figure S-18), and for 8-2, 8-3, 8-5, and 10-1 in Figure 11. For 10-3, the only isolated C60(CF3)n compound in which a fulvene fragment is isolated from the rest of the fullerene π system by six of the ten cage C(sp3) atoms (see Figure 1), the potential is only 0.17 V. In this case, the LUMO is essentially a localized fulvene-like π* molecular orbital (not shown), and this highly localized orbital leads to a relatively high E(LUMO) and a relatively low E1/2 value for this molecule. In contrast, the isomers of C60(CF3)8 and C60(CF3)10 with E1/2 > 0.30 V all have fulvene fragments that are in conjugation with the rest of the fullerene π system and consequently have LUMOs that include, to varying extents, adjacent portions of the fullerene π system in addition to the large orbital contributions from the six fulvenoid carbon atoms (hereinafter referred to as delocalized fulvene-like LUMOs). In fact, even 6-1 (E1/2 ) 0.26 V), with only six CF3 groups but with a delocalized fulvene-like LUMO, has a higher 0/- E1/2 value than 10-3. The compound 6-1 can be compared with one of its probable precursors, the hypothetical compound 4-2-CF3, which has a para-para-para ribbon of edge-sharing C6X2 hexagons and therefore has a fulvene fragment with an associated delocalized fulvene-like LUMO, as shown in Figure 11. In this case however, because the p3 ribbon is not part of a longer ribbon

ARTICLES

Figure 11. Schlegel diagrams showing the DFT-predicted cage carbon atom contributions to the LUMO of selected C60(CF3)n compounds. The blue (+) and green (-) circles represent the upper lobes of the π atomic orbitals for each cage carbon atom scaled approximately to its contribution to the LUMO. The E1/2 values shown are for the C60(CF3)n0/- couple vs the C600/couple. The E1/2 value for 4-2-CF3 was calculated using the linear leastsquares equation in Figure 9.

of edge-sharing C6X2 hexagons, as is the case with 6-1, the LUMO of 4-2-CF3 is more extensively delocalized on both sides of the fulvene fragment instead of only on one side as in the LUMO of 6-1. It is probably for this reason that 4-2-CF3, with only four CF3 groups, is predicted to have a 0/- E1/2 value (0.29 V) which is slightly more positive than the E1/2 value of 6-1, with six CF3 groups (0.26 V), as shown in Figure 9. The second isolable isomer of C60(CF3)6, 6-2,21 also has a pair of conjugated DBIPs, but in this case they each have only one C(sp2) nearest neighbor, not two. In addition, this cisbutadiene fragment is isolated from the rest of the fullerene π system. As a consequence, the LUMO is not associated with this pair of short t-DBIPs (PBE-predicted distance ) 1.358 Å21) but instead forms a localized LUMO antipodal to the skewpentagonal-prism array of CF3 groups, as shown in Figure S-12 (see SI). Furthermore, this orbital has considerably more antibonding character than does the delocalized fulvene-like LUMO of 6-1. In addition, adding an electron to the LUMO of 6-1 shortens the long C-C bonds and lengthens the short bonds in the fulvene fragment, making this fragment slightly more aromatic than the neutral compound (see Figure S-18 in SI). The combination of these factors results in E(LUMO) values J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11565

Popov et al.

ARTICLES

for this pair of compounds that differ by 0.42 eV and E1/2 values that differ by 0.33 V (with an anodic shift from C60 in the case of 6-1 and cathodic shift from C60 in the case of 6-2). The compound 12-3, with two isolated cis-butadiene LUMO fragments, is similar to 6-2 in its electrochemical behavior. A compound with two isolated cis-butadiene fragments is 104. Unlike 6-2, however, 10-4 also has two nt-DBIPs, and they form a trans-butadiene-like fragment that is in conjugation with the main π system of the molecule (the experimental and PBEpredicted distances for the symmetry-related nt-DBIPs in 10-4 are 1.365(3) and 1.376 Å, respectively). The LUMO of 10-4, shown in Figure S-10, resembles the LUMO of trans,trans,transoctatetraene. This orbital has much more antibonding character and much less bonding character than do the LUMOs of the molecules with fulvene fragments. Consequently, it is clear why 10-4 has a first-reduction potential (0.07 V) that is cathodically shifted by 0.10 V from 10-3, with its isolated fulvene LUMO, and cathodically shifted by 0.25 V or more from 10-1 and 102, with their delocalized fulvene-like LUMOs. Nevertheless, in all cases discussed so far, molecules with two conjugated ntDBIPs have LUMOs that include significant p-orbital contributions from these four cage C atoms. So far, two factors that significantly affect the E1/2 values of C60(CF3)n compounds (and probably those of other C60(X)n compounds with similar structures) have been identified: (i) compounds with a p3 ribbon fragment of edge-sharing C6X2 hexagons have LUMO fragments that resemble the LUMO of the hydrocarbon fulvene and have high E1/2 values relative to those of isomers without a p3 ribbon fragment; and (ii) the degree to which the LUMO is delocalized (i.e., conjugated to the rest of the fullerene π system) strongly affects the 0/- E1/2 value (because it affects E(LUMO); this, of course, is a general principle of any molecular orbital formalism, including Hu¨ckel MO theory). The number of substituents is not among the most influential factors. There are several compounds we have studied electrochemically that do not have two nt-DBIPs forming a fulvene fragment but instead have only one nt-DBIP. These are 2-1, 4-1, 8-4, and 10-5. The 0/- E1/2 values for 2-1 (0.11 V vs C600/-) and 10-5 (0.12 V) are the same to within experimental error, despite the large difference in the number of CF3 substituents. As far as E(LUMO) and E1/2 values are concerned, it appears that too many electron-withdrawing groups can be “too much of a good thing” in some cases. Close inspection of Figure 11 reveals that the LUMO for 2-1 is more delocalized, on the far side of the molecule (i.e., the side opposite the main part of the LUMO), than the 10-5 LUMO because 10-5 has multiple CF3 groups on the far side and 2-1 does not. The main part of the LUMO for 2-1, 4-1, 8-4, and 10-5 resembles the LUMO of acenaphthalene. Adding an electron to the LUMO of acenaphthalene or 2-1 shortens the long C-C bonds and lengthens the short ones in acenaphthalene or the acenaphthalene fragment of 2-1, making them slightly more aromatic (see Figure S-19 in SI), but to a lesser extent than was observed for fulvene or the fulvene fragment of 10-2. Consequently, which kind of LUMO fragment leads to a higher first-reduction potential, fulvene or acenaphthalene? Our results clearly show that a fulvene-like LUMO for a C60(CF3)n derivative makes that derivative easier to reduce than an acenaphthalene-like LUMO when n ) 4 (4-2 vs 4-1), when n 11566 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

) 10 (10-2 vs 10-5), and presumably for other n values, all other things being equal. Nevertheless, an acenaphthalene-like LUMO appears to have a lower energy than LUMOs that arise from 1,2 additions. This was first pointed out by Kadish et al., who reported that 2-1Bn was 0.10 V easier to reduce than 2-2-Bn62 and that 4-1-Bn was 0.05 V easier to reduce than another isomer of C60(Bn)4 with an ortho-meta-para addition pattern.57 Wudl et al. found that the first-reduction potential of 2-2-CN was 0.12 V vs C600/-,121 but the first-reduction potential of 2-1 is essentially the same relative to C600/-, 0.11 V, even though the cyano group is a much better electron-withdrawing group than the trifluoromethyl group (see Figure 6). Furthermore, the PBE-predicted first-reduction potential for 2-2-CF3 is 0.03 V, 80 mV lower than the measured value for 2-1 (and 60 mV lower than the predicted value for 2-1). Another example is the 0.140 eV difference in E(LUMO) values for 4-1-F and 4-3-F, which are -4.761 eV and -4.621 eV, respectively. Now we can return to the seven isomers of C60(CF3)8 and C60(CF3)10 with E1/2 g 0.30 V. These are 8-1, 8-2, 8-5, 10-2, and 10-6 with E1/2 ) 0.32 ( 0.01 V, and 8-3 and 10-1, with E1/2 g 0.45 V. All have a fulvene fragment with two nt-DBIPs. Why are 8-1 and 8-3 separated by 0.12 V? Why are 10-2 and 10-1 separated by 0.25 V? The simple, but incomplete, answer is that 8-1 and 10-2 only have the fulvene fragment nt-DBIPs but 8-3 and 10-1 also have another nt-DBIP (the one that is associated with the isolated p-C6(CF3)2 hexagon in the 8-3 and 10-1 addition patterns), and this provides 8-3 and 10-1 with LUMOs that have fulvene-like and acenaphthalene-like fragments that overlap. These extensively delocalized LUMOs lead to the highest 0/- E1/2 values for all C60(CF3)n compounds with n ) 2-18 listed in Table 4. The reason this answer is incomplete is because 8-2, 8-5, and 10-6, as does 8-3, have an additional nt-DBIP associated with their isolated p-C6(CF3)2 hexagons, or, in the case of 10-6, its pmp fragment. Why then are the E1/2 values for 8-2, 8-5, and 10-6 0.13 ( 0.01 V lower than for 8-3? The answer to that question is as follows. The position of some isolated p-C6(CF3)2 hexagons, like those in 8-3 and 10-1, allows the acenaphthalene-like π* fragment associated with the nt-DBIP to overlap with the fulvene-like π* fragment associated with the conjugated pair of nt-DBIPs, and when that occurs, the LUMO of the molecule includes both π* fragments, is extensively delocalized, and has a relatively low energy (see Figure 11). For 8-2, 8-5, and 10-6, on the other hand, the two π* fragments are farther apart and do not overlap (i.e., the two fragments do not contain cage carbon atoms in common). In these cases, the LUMO is essentially a delocalized fulvene-like π* orbital very similar to the LUMO of 8-1, and all three molecules have the same firstreduction potential to within experimental error. Interestingly, the LUMO+1 orbitals of 8-2, 8-5, and 10-6 are essentially an acenaphthalene-like π* orbital, much like the LUMOs of 2-1 and 4-1. In fact, the LUMO+1 energies of 8-2 and 8-5 (-4.594 and -4.666 eV, respectively) are the same to within 0.016 eV as the LUMO energies of 2-1 and 4-1 (-4.592 and -4.682 eV, respectively). The recently reported compound 12-228 has the same situation as 8-2, 8-5, and 10-6: the three nt-DBIPs give rise to a fulvenelike LUMO and an acenaphthalene-like LUMO+1 instead of combining to form one lower-energy LUMO, and consequently,

Investigation of Frontier Orbitals of Fullerene Derivatives

the first-reduction potential of 12-2 is same as the E1/2 for 8-2. The separation of the 12-2 fulvene-like π* fragment and the acenaphthalene-like π* fragment into two separate empty orbitals is clearly shown in Figure S-20 (see SI). This lower energy of the fulvene-like LUMO relative to that of the acenaphthalene-like LUMO+1 reinforces the conclusion stated earlier that, if the goal is to prepare a better one-electron acceptor fullerene derivative, addition patterns that produce fulvene-like π* fragments are more desirable than those that produce acenaphthalene-like π* fragments. It is ironic that double bonds in pentagons, which are destabilizing as far as the total energy of a fullerene derivative is concerned,52,53 are exactly what is needed to lower the energy of its LUMO and increase its electron affinity. However, DBIPs are necessary but not sufficient. Most C60(CF3)n compounds have numerous t-DBIPs (i.e., the blue DBIPs in Figure S-8), and these are among the shortest C-C bonds in the cage (see the X-ray structures cited in Table 1). None of the LUMOs we have examined are ever associated with t-DBIPs. Only nt-DBIPs with two C(sp2) nearest neighbors generate low-energy π* fragments that are incorporated into the LUMO. The compound 12-1, for example, has six t-DBIPs and twelve electron-withdrawing CF3 groups but is 0.16 V more difficult to reduce than C60 (this addition pattern was previously predicted by Dixon et al.53 to lead to high E(LUMO) values relative to C60, even for C60F12). We have predicted the 0/- E1/2 values for the recently reported compounds 16-1, 16-2, 16-3, 18-1 and for the hypothetical compound 18-2, and they are listed in Table 4. The LUMOs for four of these compounds have one or more fulvene-like π* fragments, none of which overlap with each other; the LUMO for 16-2 does not resemble any obvious aromatic hydrocarbon fragment (all five LUMOs are shown as Schlegel diagrams in Figure S-21). None of these compounds is predicted to be easier to reduce than 8-3. The compound 16-1 has an E1/2 value of only 0.04 V vs C600/-. The highest predicted E1/2 value for these five compounds is for 18-2 (0.36 V vs C600/-), which has a LUMO with three isolated fulvene π* fragments. The concept that the overlap of two π* fragments can combine to produce a low-energy LUMO led to the search for an addition pattern with a higher E1/2 value than 10-1. The C2symmetry compound 8-7-CF3, which has two separate p3 ribbons of edge-sharing p-C6(CF3)2 hexagons, as shown in Figure S-22, has a LUMO in which the two fulvene-like π* fragments are efficiently overlapped and has predicted E(LUMO) and 0/- E1/2 values that are 1.000 eV lower and 0.73 V higher, respectively, than that of C60. This hypothetical isomer of C60(CF3)8 is only 13 kJ higher in energy than 8-2, the most stable isomer of this composition (see Table S-2 in SI). In summary, for the numerous addition patterns we have investigated, the electronic properties of substituents as diverse as CH3 and CN can produce variations in C60(X)n E1/2 values of more than 3 V. The variation depends on the value of n but is almost independent of the addition pattern. For a given substituent, the range of experimentally observed E1/2 values depends on n, but the range for a set of isomers can be so large (e.g., up to 0.5 V for n ) 10) that it is essentially the addition pattern, not the value of n, that determines the particular E1/2 value for a given compound. From the standpoint of designing better electron acceptors, structures with nt-DBIPs are essential,

ARTICLES

and therefore 1,4 additions are better than 1,2 additions. Furthermore, (i) addition patterns that produce a delocalized fulvene-like π* fragment are better than those that produce an acenaphthalene-like π* fragment, (ii) addition patterns that result in the overlap of fulvene-like and acenaphthalene-like π* fragments are even better yet, and (iii) addition patterns that result in the overlap of two fulvene-like π* fragments are best. V. Correlations Involving -/2- and 2-/3- E1/2 Values. Sixteen second reductions and ten third reductions were reversible (as defined earlier). Although the second added electron should, in principle, be paired with the first added electron in the LUMO (unlike C60, which has a triply degenerate LUMO, all 16 C60(CF3)n compounds with a reversible second reduction have Cs, C2, C2h, or C1 symmetry and therefore cannot have degenerate orbitals), there was a poor correlation between the E(LUMO) and -/2- E1/2 values. There was also a poor correlation between the E(LUMO+1) and 2-/3- E1/2 values. Since the poor correlations might be due to (i) the increasing solvation energies of the C60(CF3)nm- anions as m increases and (ii) structural and electronic reorganization following electron transfer, we decided to compute the energies of the C60(CF3)nmand C60(CF3)n(m+1)- anions in the gas phase as well as their solvation free energies in dichloromethane and to model the reduction potential as the difference in energy between the solvated C60(CF3)nm- and C60(CF3)n(m+1)- anions. The electrostatic contribution to the solvation free energies were calculated using the Born model (BM) and the conductorlike polarizable model (C-PCM; see Experimental Section). The effective radius of each ion used for the BM calculations was the cube root of 75% of the C-PCM cavity volume. The computed solvation energies, which decreased as n increased (because of the increase in volume as n increased), were similar for both models. The C-PCM results are listed in Table S-3 along with the B3LYP-predicted electron affinities and reduction potentials. (The B3LYP-predicted EA(1) values for C60 and C70, 2.519 and 2.594 eV, respectively, are only ca. 0.15 eV lower than the experimental values, 2.666(1) eV)4,5 and 2.676(1) eV,4 respectively.) Note that the electrostatic solvation energies of the neutral compounds are all less than 0.1 eV. As expected, the predicted EA(1) values for C60(CF3)n derivatives are higher than for C60. The values for 4-1 (2.961 eV), 6-1 (3.145 eV), the five isomers of C60(CF3)8 (2.9613.448 eV), and the five isomers of C60(CF3)10 (3.010-3.612 eV), compare favorably with the experimental EA(1) values for samples of C60(CF3)4-10 consisting of an unknown mixture of isomers, which are 3.03 ( 0.20 eV125 and 3.17 ( 0.19126 for C60(CF3)4, 3.04 ( 0.14125 and 3.17 ( 0.16 eV126 for C60(CF3)6, 3.07 ( 0.06125 and 3.12 ( 0.18 eV126 for C60(CF3)8, and 3.200 or 3.245 eV for C60(CF3)10.125 However, the experimental values are virtually the same to within the quoted errors, but our calculations show that the ranges of EA(1) values for a set of isomers can be as large as 0.6 eV. The B3LYP-predicted EA(2) values for some of the C60(CF3)n compounds are positive, which indicates that their dianions should be stable species in the gas phase. Indeed, C60(CF3)n2(125) Markov, V. Y.; Aleshina, V. E.; Borschevskii, A. Y.; Khatymov, R. V.; Tuktarov, R. F.; Pogulay, A. V.; Maximov, A. L.; Kardashev, S. V.; Ioffe, I. N.; Avdoshenko, S. M.; Dorozhkin, E. I.; Goryunkov, A. A.; Ignat’eva, D. V.; Gruzinskaya, N. I.; Sidorov, L. N. Int. J. Mass Spectrom. 2006, 251, 16-22. (126) Borschevskii, A. Y.; Aleshina, V. E.; Markov, V. Y.; Dorozhkin, E. I.; Sidorov, L. N. Inorg. Mater. 2005, 41, 1318-1326. J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007 11567

Popov et al.

ARTICLES

(ca. 8-10 eV), and relatively small uncertainties in anion volumes lead to larger errors in E1/2 values for the third reductions than for the first and second reductions. The connection between addition pattern and E(LUMO) discussed in the previous section should prove useful in understanding gas-phase electron affinities (EAs) as well as solution E1/2 values. In addition to their fundamental importance, reliable electron affinities of fullerene(X)n compounds, whether determined experimentally or predicted with a validated computational method, can be used to estimate negative ion chemical- or electrospray-ionization sensitivities, which can be useful for the compositional analysis of reaction mixtures by mass spectrometry.128 Summary and Conclusions

Figure 12. Plots of experimental second-reduction potentials of 18 C60(CF3)n compounds (relative to the C60-/2- potential) vs the B3LYP-predicted relative second reduction potentials relative to the predicted C60-/2- potential (blue points) or the B3LYP-predicted second gas-phase electron affinities relative to the predicted EA(2) for C60 (red points).

(n ) 2-20) dianions were generated recently by collisions of the corresponding monoanions produced in an electrospray ion source with sodium atoms.127 Cross sections for electron capture by monoanions decreased on going from n ) 10 to n ) 20. Figure 12 displays plots of 17 experimental -/2- E1/2 values (including C60-/2-) vs (i) B3LYP-predicted EA(2) values (relative to C60) and (ii) B3LYP-predicted -/2- E1/2 values (i.e., the difference in ∆Gsolv of C60(CF3)n(m+1)- and C60(CF3)nmminus the EA(m+1) value, also relative to C60). The correlation between the experimental E1/2 values and EA(2) is poor (R2 ) 0.50), but when corrected for the differences in ∆Gsolv for the mono- and dianions, the correlation is sufficiently better (R2 ) 0.94) so that -/2- E1/2 values predicted for C60(CF3)n compounds not yet examined by cyclic voltammetry are probably reliable to (0.04 V. Note that there are no points for 8-5 and 12-1 in Figure 12 because these compounds do not have reversible second reductions. As expected, solvation corrections also improved the correlation between experimental 0/- E1/2 values and B3LYPpredicted EA(1) values (not shown). However, even when solvation was taken into account, the B3LYP-predicted thirdreduction potentials were poorly correlated with the experimental values. This may be because the electrostatic contributions to the solvation energies for the C60(CF3)n3- trianions are very large (127) Streletskiy, A. V.; Hvelplund, P.; Nielsen, S. B.; Liu, B.; Tomita, A.; Goryunkov, A. A.; Boltalina, O. V. J. Chem. Phys. 2006, 124, 144306144316. (128) Betowski, L. D.; Enlow, M.; Aue, D. H. Int. J. Mass Spectrom. 2006, 255-256, 123-129.

11568 J. AM. CHEM. SOC.

9

VOL. 129, NO. 37, 2007

Experimental reduction potentials for 18 C60(CF3)n compounds have demonstrated that the addition pattern is as important, if not more important in many cases, than the number of substituents, n, in determining E1/2 values. DFT calculations demonstrate that those addition patterns with double bonds in pentagons having two C(sp2) neighbors result in the strongest electron acceptors. Specifically, (i) compounds with a delocalized fulvene-like π* fragment are better electron acceptors than those with an acenaphthalene-like π* fragment, (ii) compounds with addition patterns that result in the overlap of fulvene-like and acenaphthalene-like π* fragments are better yet, and (iii) compounds with addition patterns that result in the overlap of two fulvene-like π* fragments are predicted to be the best electron acceptors. Acknowledgment. We dedicate this paper to Neil Bartlett on the occasion of his 75th birthday. He has always been and continues to be both a friend and an inspiration to us and many others in the field of fluorine chemistry. We appreciate the financial support of the DAAD, the Volkswagen Foundation (I-77/855), the Fonds der Chemischen Industrie, Germany, the Russian Foundation for Basic Research (06-03-33147-a), the Russian Federal Agency of Science and Innovations (Contract 02.442.11.7439), the Civilian Research and Development Foundation (RUC2-2830-MO-06), the U.S. National Science Foundation (CHE-0707223 and, for partial support of EBS, CHE0416017), and the Colorado State University Research Foundation. We thank Karl F. Freed for encouragement and assistance, Manfred Kappes for generous support, Donald L. Dick for assistance with APCI mass spectrometry, Xiang Gao for digitized absorption spectra of the two isomers of C60(Bn)4, and Karl Kadish, Carlo Thilgen, and Warren Powell for helpful discussions. Supporting Information Available: Crystallographic data in CIF format, complete references 40 and 73, and additional figures and tables. This material is available free of charge via the Internet at http://pubs.acs.org. JA073181E